Home

LCModel & LCMgui User`s Manual - s

1. With PV5 you should be able to do the following Set the parameter PVM_RefScanYN to Yes The unsuppressed water reference is then in the file fid refscan In the file method PVM_RefScanNA NSunsuppressed PVM_RefScanRG Gainunsuppressed In the file acqp NA N Ssuppressed RG Gainsuppressed 10 2 WATER SCALING 131 10 2 2 3 Partial Volume Correction Many do not bother with this The partial volume correction is often small and the segmentation of the VOI into classes is seldom accurately known You may at least want to consistently assume the same segmentation unless you have clear knowledge of a different segmentation Suppose you segment into 3 classes fym fum and fess with fom fum fest 1 where fess is the volume fraction of the VOI with a nearly pure water signal Then instead of WCONC suggested in Sec 10 2 2 you would input 43300 fom 35880 fum 55556 fos f WCONC t fest The numerator is simply the total water concentration in the VOI The additional factor 1 1 fesf is not part of the water concentration it is the partial volume correction It corrects for the fact that the metabolites are practically all concentrated in the gray matter and white matter You can see from the equations in Sec 10 2 1 that WCONC multiplies fscale and therefore the absolute concentrations output will then already have the partial volume correction Chapter 11 Fine Points This chapter contains a collection of f
2. Cr REAL the chemical shift in ppm of the singlet used for scaling Default WSPPM 3 027 INTEGER the number of equivalent protons contributing to the singlet used for scaling Default N1HMET 3 130 CHAPTER 10 ABSOLUTE METABOLITE CONCENTRATIONS 10 2 2 1 Bruker ParaVision 6 Data PV6 0 is missing information for Water Scaling use PV6 0 1 or higher LCMgui does not divide the Bruker data by the total gain Gain With Water Scaling you must multiply the LCModel concentrations by the factor Gainunsuppressed Gadinsuppressed Most convenient is to include this correction in LCModel by inputting ATTH20 multiplied by this factor Then the output concentrations are already multiplied by this factor The default value of ATTH20 is 0 7 so for example if the above factor were 0 2 you would input ATTH20 0 14 In the file method Gainunsuppressed PVM_RefScanRG In the file acqp Do not use RG as in older PV versions Instead Gainsuppressed the 5th value in the line following ACQ_jobs 10 2 2 2 Bruker ParaVision 5 Data and lower LCMgui does not divide the Bruker data by the number of scans NS or by the total gain Gain With Water Scaling you must multiply the LCModel concentrations by the factor N Sunsuppressed N Ssuppressed x Gainunsuppressed GQiNsuppressed More convenient is to include this correction factor in LCModel by inputting ATTH20 multiplied by this factor as above in Sec 10 2 2 1
3. ECC ERRMES 11 FATAL All channels or spectra to be averaged have been eliminated from the analysis Hint Check your input of NDCOLS ICOLST ICOLEN NDROWS IROWST IROWEN NDSLIC ISLICE NVOXSK ICOLSK or IROWSK evaluates the incomplete beta function which is used to compute the confidence regions 11 for ag and ag in 1 Eq 4 1 FATAL meaningless arguments in call to BETAIN probably caused by more than 40 000 data points Hints Never Happened 2 ERROR A series expansion has failed to converge ag and ag could be poorly chosen Hints Never Happened processes the CHCOMB Control Parameters CHARACTER 100 210 these must satisfy the rules in Sec 9 9 1 1 The Detailed Output ends with Incorrect CHCOMB followed by the first 112 characters of the CHCOMB Hints Namelist errors are the most common cause e g forgetting a single quote for a CHARACTER Control Parameter somewhere in the Namelist 1 FATAL NCOMBI exceeds 200 2 FATAL A Metabolite Name in a CHCOMB has more than 6 characters 3 FATAL A CHCOMB has more than 50 Metabolite Names computes the initial approximations for the priors for the concentration ratios See Sec 11 8 for the symbols and tokens used below 1 FATAL NRATIO must not exceed 100 2 FATAL In one of the CHRATO the is missing in metnum metdenom The illegal ratio part of the CHRATO is also at the end of the Detailed Output 3 FATAL In one of the CHRATO the metnum is missing in
4. cd lcmodel bin plotraw lt 1lcmodel test control plotraw in This produces the output file HOME 1cmodel test output plotraw ps which you can send to your PostScript printer It contains plots of the test data in HOME 1cmodel test raw test RAW on three pages the time domain data the frequency domain data and the absolute value of the frequency domain data 8 5 2 PlotRaw IN File PlotRaw needs two input files RAW file the identical file with the raw time domain data that all the other programs use it is specified in Sec 5 2 IN file the file with the changes to the default parameters that control the plot As you can see from the example plotraw in the IN file consists of only the Namelist PLTRAW You must input the following in Namelist PLTRAW HZPPPM NUNFIL DELTAT Same as in Sec 5 3 1 FILRAW CHARACTER 255 the name of the RAW file containing the data to be plotted FILPS CHARACTER 255 the name of the PostScript output file that will contain the plots You will want to input values for the following unless you want the default values to be used 88 PLTIME PLFREQ PLABSV DEGZER DEGPPM PGNORM PPMST PPMEND CHAPTER 8 MAKING THE BASIS SET LOGICAL TRUE to plot the real and imaginary parts of the time domain data on one page Default PLTIME T LOGICAL TRUE to plot the real and imaginary parts of the frequency domain data on one page Default PLFREQ T
5. 02 FWHM 17 lt 2 02 AMP 2 CHSIMU 11 MM17 1 67 03 FWHM 15 lt 17 02 AMP 2 CHSIMU 12 CrCH2 3 93 0 FWHM 9 lt 0 0 AMP 2 CHSIMU 13 Gua 3 78 0 FWHM 9 lt 0 0 AMP 2 CHSIMU 14 Glyc 3 55 0 FWHM 9 lt 0 0 AMP 2 These are of the form metabo ppmi sdppm FWHM fwmin1 lt fwex sdfw AMP ampl O ppm2 FWHM fwmin2 AMP amp2 metabo the simulated Metabolite Name ppm the expectation ppm value of peak There can be up to 20 peaks sdppm the positive standard deviation of the shift in the simulated spectrum fwmin the minimum Gaussian FWHM of peak fwex the expectation FWHM of peak 1 which is exponentially Lorentzian broadened All peaks in this Basis Spectrum get equal exponential broaden ing You must input fuminl lt fwex sdfw the positive standard deviation of the expectation fwex amp the effective number of protons contributing to peak The amplitudes are scaled consistently with the spectra in the BASIS file using WSMET WSPPM amp N1HMET as in Sec 10 2 2 So if the spectra in the BASIS file are scaled to yield mM then AMP 2 0 would yield mM of CH2 groups not necessarily mM of molecules fwmini lt O simulates a sum of singlets with the same FWHMBA as the Basis Spectra in the BASIS file and with same restricted broadening as with CrCH2 amp Glyc above In this case only the inp
6. CHNOT2 3 Lip13e These are only needed to give more flexibility to the Preliminary Analysis which uses more primitive models than the Final Analysis HIFMM REAL High field MM When HZPPPM gt HIFMM sharper CHSIMU for the MM amp lipids are used with FWHM about 1 3 those of above Default HIFMM 190 11 7 1 Method Lipid and macromolecule MM signals are highly variable and it is therefore difficult to spec ify accurate Basis Spectra for them in contrast to the usual low molecular weight metabolites in the BASIS file When there is uncertainty in the model constrained regularization 11 which is nearly model free can be useful Figure 4 in 16 shows how the regularized LCModel Baseline can automatically account well for a strong MM signal in high quality data with no prior knowledge of or biasing assumptions about the MM signal However with lower resolution and S N the information is not sufficient to determine such sharp high amplitude bends in the Baseline which occur in high resolution spectra Three approaches have been useful with LCModel 1 Add an extra typical MM spectrum obtained from metabolite nulling to the Basis Set 16 18 Even if this MM Basis Spectrum does not perfectly match each in vivo spectrum the Baseline now only needs to make smaller corrections which are easier than fitting the full MM profile especially with lower quality data 2 A metabolite nulled spectrum can be
7. Data Conversion A program is executed to produce the LCModel RAW file and any available Control Parameters e g NUNFIL DELTAT from the file selected in Step 1 You can specify this program to be the standard LCMgui program for the standard Bruker GE Marconi Picker Philips Siemens Toshiba or Varian files Sec 3 7 However there is also an Other button to use a script supplied by you for another type of file Preprocessor Even if you specify one of the standard programs for Step 2 you can still supply your own script in this step to modify the RAW file or Control Parameters from Step 2 The original default when LCMgui is installed is to use no Preprocessor Control Parameters Window This window Fig 7 2 allows you to make many checks of the data and modifications of the Control Parameters and the LCMgui configuration before starting LCModel You can also configure LCMgui so that the Normal User usually only has to click Run LCModel at the bottom of the window without much knowledge of the rest Run LCModel When you click the Run LCModel button LCMgui makes many checks and then starts an Execution Script in background The default script starts LCModel and archives the results However this script is available to you as a template and you can modify it for your own 53 54 CHAPTER 7 LCMGUI REFERENCE MANUAL purposes The possibility of supplying your own scripts in Steps 2 3 amp 5
8. Eddy Current correction Sec 8 4 Run MakeBasis three times with PPMST PPMEND 8 54 8 34 4 2 0 2 0 1 0 1 to evaluate the resolution and quality also of the formate and DSS markers Sec 8 6 2 Run MakeBasis three times with PPMST PPMEND 2 1 1 9 3 1 2 9 3 3 3 1 to estimate FWHMBA which you now input in Namelist SEQPAR in your makebasis in Sec 8 6 4 1 Calibrate GPC PCh and NAAG NAA Sec 8 7 1 In your makebasis in correct CONC for GPC and if necessary for NAAG Run MakeBasis and repeat the calibration runs The GPC and NAAG concentrations output by LCModel should now be 1 0 Then your Basis Set should be ready for use Chapter 9 Further Useful Options and Information This chapter specifies useful Control Parameters that you can input in the CONTROL file or with LCMgui Fig 7 7 otherwise they remain at their default values You should at least scan over the options in this chapter 9 1 Special Types of Spectra LCModel uses the choline creatine amp NAA landmarks at 3 22 3 03 amp 2 01 ppm for initial referencing You can inform LCModel of missing landmarks or of special types of spectra remember that PLOTs are in HOME 1cmodel1 doc figures pdf SPTYPE CHARACTER 16 special type SPTYPE muscle 5 for muscle spectra PLOT 8 To use this you must read Sec 9 2 SPTYPE lipid 8 for spectra with only lipid and water signals PLOT 9 shows a vertebra
9. HZPPPM 85 15 to roughly account for the decrease in T2 with field strength For each of the above three quantities there are four types of Control Parameters These four are listed below where can be EXT2 SDT2 or SDSH DE REAL default these specify the defaults for the three quantities for all metabo lites They will be used unless they are changed by the Control Parameters below Defaults DEEXT2 2 0 DESDT2 0 4 DESDSH 0 004 N INTEGER number the number of quantities that are to be changed from their defaults Defaults NEXT2 0 NSDT2 0 NSDSH 3 CH CHARACTER 60 6 character the first N elements contain the Metabo lite Names corresponding to the quantities to be changed Defaults CHSDSH 1 NAA CHSDSH 2 NAAG CHSDSH 3 Cho AL REAL 60 alternate the first N elements contain the changes to the quan tities Defaults ALSDSH 1 0 002 ALSDSH 2 0 002 ALSDSH 3 0 008 As explained in 1 Appendix the uncertainties in the referencing of NAA and NAAG are reduced to prevent them from strongly overlapping The uncertainty in the referencing of Cho is increased because the Cho chemical shift does not agree with those of the choline containing compounds in vivo It is better to use GPC amp PCh instead of Cho The lineshape parameters can automatically account for increases in 1 T2 from in vitro to in vivo The parameters in the priors are needed to
10. essary 23 Water Scaling Secs 6 3 4 amp 10 2 is a convenient alternative for all cases 3 7 8 Toshiba LCMgui handles rawData files V4 10 V6 30 for both 2D CSI amp single voxel data 3 7 VENDOR SPECIFIC GUIDELINES 27 LCMgui handles DICOM files V7 xx and later for both 2D CSI amp single voxel data With CSI data the unsuppressed water reference must have the same number of voxels as the water suppressed scan i e you must set ref_mode to 1 not 11 LCMgui only acccepts CSI single channel not CSI multi channel files With single voxel spectra you should select Do eddy current correction in LCMgui Sec 6 3 3 3 7 9 Varian Agilent 3 7 10 LCMgui handles single channel fid files CSI amp single voxel It handles multi channel single voxel but not CSI fid files The corresponding file procpar must be in the same directory with fid With single voxel data or multi channel data you should select Do eddy current correction in LCMgui You will then be prompted to select the unsuppressed water reference fid file corresponding to your selected data file With multi channel data you must input IAVERG T Sec 11 9 2 With single voxel single channel spectra you can increase the speed of the analysis by using only the first 2 complex data points where N is the largest possible integer You only have to input this as NUNFIL Sec 5 3 1 you do not have to change your file fid Other Vendors
11. info An extra iteration was required in determining the phases Hints Working Hard info Extra iterations were required in determining the phases Hint This is common in spectra dominated by lipids as muscle liver bone marrow etc Imposing prior knowledge about phase corrections Sec 9 7 can help greatly warning A regularized solution within the specified confidence region could not be found at some stage The closest solution will be used Hint This is fairly common in spectra dominated by lipids as muscle liver bone marrow etc If it does not occur in the final stage the solution will hardly be affected Not used info A more detailed search for phase corrections will be made This can increase the computation time considerably Hint This is common in spectra dominated by lipids as muscle liver bone marrow etc FATAL DDEGP3 is non positive or MDEGP3 lt 2 Hint Do not input these Hidden Control Parameters warning Too many extra searches from a grid of first order phase correc tions are needed The grid will be made coarser Hints An extreme first order phase correction may be needed If so you should input DEGPPM DGPPMN and DGPPMX to specify its range If you have already input a wide range with DGPPMN amp DGPPMX then there is nothing further to do Not used info The usual solution found a first order phase correction outside the range of DGPPMN and DGPPMX A further search for better phas
12. 11 11 Analyzing Data Left amp Right of the Water Signal 11 12 Multimodal Lineshapes 0 0 0 00000 000 5 10 CONTENTS 11 13 Relaxation amp Shift Priors 2 2 2 2 ee ee 148 1114 Detailed Output oli ke ese Oke eae fe ge eh ee 149 11 15 Default Control Parameters 2 2200 0004 150 11 16 Analyzing Magnitude Spectra e 151 11 17 Consistency with Old Versions 02 0200 151 11 18 Nuclei Other than H hl kde lt Bote boas ek Ba 152 12 Diagnostics and Troubleshooting Hints 153 12 1 Non Standard Diagnostics 2 2 0 0 00000000 2a ee 153 IZED T O Errors oe ES AS eS ee Se le ee ES 153 12 1 2 Arithmetic Exceptions 2 020004 154 12 1 3 Other Aborts with no One Page Output 154 12 2 Standard LCModel Diagnostics o e e e 154 Chapter 1 Preface amp Overview The LCModel package is for the automatic quantitation of in vivo proton MR spec tra 1 LCMgui is a graphical user interface for running LCModel with more con venience and less knowledge of this manual LCMgui is provided without charge for use with LCModel only 1 1 What You Must Read In order to avoid the proliferation of documents this is the only manual for the LCModel package and LCMgui It is a detailed reference manual However as outlined below most of you can use this as a shorter users guide consisting of only a few of the chapters here In addition
13. CHUSE1 CHARACTER 300 6 the first NUSE1 elements contain the Metabolite Names of the spectra to be used Default CHUSE1 NAA Cr GPC Glu Ins ECHOT REAL The echo time in ms If ECHOT gt 100 0 then the default becomes NUSE1 4 because of the weak Ins at long TE However you can override this by simply inputting your own value for NUSE1 11 6 Unusual Phantoms By default LCModel assumes that the spectrum being analyzed contains dominating land mark resonances from NAA Cr or Cho These are used to get a starting point for referencing A unusual phantom is one that has none of these landmarks dominating You would never use such a phantom for calibration With such a phantom you must use the Control Param eters in the following three steps 1 VITRO T 2 You must also help with the referencing There are two possibilities 2a Suppose your phantom spectrum has a strong isolated resonance say at 3 33 ppm that can be used for referencing You then input the Control Parameters NREFPK 1 PPMREF 1 2 3 33 2b Method 2a could not be used if your phantom only had metabolites like Asp GABA Glc Gln Glu or Tau In this case you must plot the spectrum with LCMgui s Preview Data Sec 7 4 1 and use PPMSHF as described in Sec 11 3 3 3 Suppose your unusual phantom contains metabolites with the Metabolite Names Metab1 Metab2 Metab3 There are two possibilities 3a NUSE1 3 CHUSE
14. Do Not Save to continue with this analysis using your changes but not saving them Save Use Next Time Do Not Save bn2 tumor Figure 7 9 Here you enter a name for the file which will store your Control Parameters for optional future use You are free to choose any descriptive filename However if it is identical with one of the filenames listed at the bottom e g bm10 then your file will overwrite the old file If you click Use Next Time in Fig 7 9 then this set of Control Parameters will be used as defaults for future sessions LCMgui will still reset obviously variable Control Parameters such as TITLE FILRAW etc In addition LCMgui may reset some Control Parameters e g SDDEGZ amp SDDEGP to their commonly used values The only way to avoid having to manually set them back to their desired values every time is to use a Preprocessor Sec 7 4 5 If you click Save then your Control Parameters will simply be archived for optional use later They will not be used as defaults for the next session The old Control 7 3 BASIC SETTINGS AND USAGE 67 l r Select Control Defaults file A Below is a list of Control Defaults files i e files with changes to Control Parameters that you may have saved from previous sessions You can choose to have one of these Control Defaults files provide the starting default values for this present session and future sessions In this way you do not have to
15. Do not input the hidden Control Parameter NBCKMN FATAL Your Analysis Window has zero width Hints This is actually ILLOGICAL since this error should have been caught earlier ILLOGICAL There are less than 6 knots for the Baseline This error should have been caught earlier 160 GETPHA GFIELD HEX ICYCLE INITIA LDEGMX CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS Get phases Phase the unsuppressed water peak 1 FATAL Part of the unsuppressed water peak is at the edge of the spectrum Hint Increase your bandwidth parses individual elements of the strings in CHSIMU amp CHRATO 1 ILLOGICAL Programming error 2 ILLOGICAL Programming error converts to hexadecimal format for the PS file 1 ILLOGICAL Programming error in this conversion No One Page Output can be produced computes subscripts in arrays from circular FFTs 1 FATAL A non positive subscript has been computed probably due to my programming error Hints Never Happened does initial computations of basic quantities Hints Errors here are almost always due to your input of unreasonable Control Parameters 1 FATAL At least one of the following Control Parameters is not positive DELTAT NUNFIL HZPPPM 2 FATAL At least one of the Control Parameters in INITIA 1 is unreasonably large Hints This is most likely NUNFIL which cannot exceed 65536 in most ver sions of LCModel Rerun to get the Detailed Output a
16. Hints In FILH20 check that the time data correspond to FMTDAT Be sure that FILH20 corresponds exactly to FILRAW except for water suppression especially for multi voxel data sets 11 warning At least one of the following input variables in Namelist NMID in your H20 file is not input positive TRAMP VOLUME The values from your RAW file will be used If these values are not the same for H20 amp RAW then this would cause an error in your Water Scaling and therefore the absolute concentrations 12 FATAL No H20 file could be read although UNSUPR T says to analyze the H20 file as data ONEPAG produces the One Page Output Aborts in ONEPAG will suppress the One Page Output 1 ILLOGICAL Meaningless argument in call to ONEPAG 2 FATAL At least one of the following Control Parameters is not positive PTLABL PTTITL PTOUTP or at least one of the following is less than 1 0 RHLABL RHTITL RHOUTP RHVERS 3 ILLOGICAL As ONEPAG 1 166 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS 4 FATAL You have allocated so much space for margins titles and labels that there is no room left for the plot or you have input negative values for XLEFT or YBOTT 5 ERROR You have input a non positive value for XSTEP It will be set to 0 20 6 FATAL The ppm axis consists of at most two Grid Points Hints Are PPMST and PPMEND very close together 7 ERROR You have input a non positive value for NSUBTK It will be set to 1 8 FATA
17. LOGICAL TRUE to plot the absolute value of the frequency domain data on one page Default PLABSV T REAL Ho your zero order phase correction in degrees e g as output by LCModel and MakeBasis This is with respect to water see the following definition of DEGPPM If PLFREQ F then you do not need to input DEGZER or DEGPPM Default DEGZER 0 REAL 1 your first order phase correction in degrees ppm e g as out put by LCModel and MakeBasis Thus the FFT of the data in the RAW file will be multiplied by exptildo 4 65 1 7 180 where 6 is the chemical shift in ppm You can quickly see from the plots of the phased spectra from PlotRaw whether our conventions for dy and Q agree For Bruker users DEGZER PHC1 2 PHCO DEGPPM PHC1 bandwidth in ppm where PHCO and PHC1 are output by the Bruker routine pk Default DEGPPM 0 Same as in Sec 5 3 6 Same as in Sec 5 3 3 except that it sets the left edge of the plotting window i e the ppm range over which the frequency domain data will be plotted The time domain data are always plotted over the full time range Default PPMST 999 i e the ppm axis will extend up to the maximum ppm value of the data The axis will only extend far enough to plot the data and not to this excessive value Same as in Sec 5 3 3 except that it sets the right edge of the plotting window Default PPMEND 999 i e the ppm axis will extend to the
18. NTITLE 2 Any value other than 1 will allow the TITLE to range over a maximum of two lines 152 CHAPTER 11 FINE POINTS YEAR4D LOGICAL year with 4 digits The Sun SGI amp DEC versions now have a 4 digit year at the top of the One Page Output Use FALSE to keep the old 2 digit year Default YEAR4D T CONC3F LOGICAL concentration to 3 figures When the concentrations in the One Page Output are in exponential E format they now have 3 significant figures Use FALSE to keep the old 2 significant figures Default CONC3F T 11 18 Nuclei Other than H LCModel was designed from the beginning for proton MRS and is restricted to this Pierre Gilles Henry 34 using many special LCModel settings has managed to analyze 13C spectra quite successfully However there are no plans to extend or support LCModel analysis of any nucleus other than H Chapter 12 Diagnostics and Troubleshooting Hints LCModel performs extensive checks for abnormal conditions and has about 300 di agnostic messages that go into the diagnostics table in the One Page Output These standard LCModel diagnostics are fully covered in Sec 12 2 Sometimes there is an abort before there is a chance for a standard diagnostic particularly with input errors These are discussed in Sec 12 1 12 1 Non Standard Diagnostics These often involve an abort with no One Page Output and therefore no plot of the data However there is usually an
19. Same as in Sec 8 5 2 except that the default is PPMEND 0 2 to see the DSS marker Note that if you input PPMST lt PPMEND then no plots will be made You probably will not have to input values for the following FMTBAS NAIM HWDSCA HWDPHA CHARACTER 80 format Basis the Fortran FORMAT specification for the Basis Spec tra output in the BASIS file Default 6E13 5 You can customize the plots e g for slides or figures by inputting values for any of the Control Variables in Secs 9 9 3 2 9 9 3 4 except PTOUTP RHOUTP RGBLIN SUBBAS and RHVERS INTEGER Same as in Sec 8 5 2 except that there is only one element in NAIM not three since MakeBasis only plots the frequency domain data Default NAIM 18 REAL half width for scaling When AUTOSC T the real part of standard formate peak will be integrated over an interval up to HWDSCA ppm about the peak maximum The default is HWDSCA 0 15 If any of your formate peaks are so broad that they have not decayed into the Baseline within this interval then you should repeat your acquisitions You can check this by running with PPMST 8 6 and PPMEND 8 3 MakeBasis effec tively reduces HWDSCA if possible so you should not have to reduce HWDSCA or HWDPHA REAL half width for phasing When AUTOPH T the phase of the real part of the DSS peak will be optimized over an interval up to HWDPHA ppm about the peak maximum The default i
20. You may choose any descriptive 7 4 FURTHER USEFUL SETTINGS 69 Select User Profile ZA Below is a list of your User Profiles The selected User Profile is shown in green right below Selected Profile You can select another User Profile by clicking on its name below Click on Use Selected Profile below when you are satisfied with the green Profile Selected Profile Figure 7 12 At the start of a session you select the User Profile to be used directory name However if it is identical with one of the names listed at the bottom e g 1 then your default files will overwrite the old files If you enter toshiba as your User Profile then your gui defaults will be in HOME 1cmodel profiles toshiba gui defaults and all Control Defaults files that you save from this User Profile will be in the directory HOME 1cmodel profiles toshiba control defaults You can look at their contents with the usual Unix commands but do not edit them You can delete obsolete Control Defaults files e g with rm HOME 1cmodel profiles toshiba control defaults bm10 Your initial User Profile is 1 You can rename this or any of your others by a command like mv HOME 1lcmodel profiles 1 HOME 1cmodel profiles siemens When you start LCMgui Fig 7 12 allows you to select any one of your User Profiles for that session if you have more than one User Profile from previous sessions 7 4 Further Useful Settings 7 4 1 Previewing D
21. and all the others are black REAL 2 2 DSHPAT JDASH JCURVE specifies the dashed pattern for the fine grid lines when JCURVE 1 and for the data when JCURVE 2 The length of the dashes is given by DSHPAT 1 JCURVE and the space between the dashes by DSHPAT 2 JCURVE If either of the lengths is non positive then a solid curve is drawn Defaults DSHPAT 05 1 2 0 LOGICAL subtract Baseline SUBBAS T subtracts the Baseline from the data and the fit before plotting them This can help visualizing the fit if the water suppression is very weak and the Baseline is much larger than the metabolite peaks Default SUBBAS F ppm Axis REAL the positive increment between the numbers on the ppm axis Default XSTEP 0 2 INTEGER the number of fine grid lines in going from one number to the next on the ppm axis Default NSUBTK 2 9 9 4 Miscellaneous Output Table More information can be output in this table I used this during development and testing IETCOU INTEGER maximum number of lines that will be output in this table Default IETCOU 3 The first three lines have already been described in Sec 2 4 The further lines in this table are 4 5 alphaB S ag and as in 1 Eq 4 spline knots Np in 1 Eq 1 Ns The number in parentheses is dn the increment in n in the sum in 1 Eq 1 The sum runs over n n Ns Ns 1 Ns 1 Ns Except for broad line shapes or ex
22. e For CSI data sets only one voxel near the center will be selected Sec tion 4 2 5 allows you to display whole slices from 2D amp 3D data sets 4 2 5 More Complete Test of LCMgui with a False License You can test all facilities of LCMgui by installing a false license e First make a false basis file with the commands mkdir p HOME 1cmodel basis sets touch HOME 1cmodel basis sets dummy basis 4 3 TEST RUNS WITHOUT LCMGUI 31 e With Linux simply enter touch HOME 1cmodel license With Sun SGI or DEC Compada install the license Sec 7 2 5 by entering any name next to License OWNER and an integer 1 9 digits long next to License KEY e You can then select one of your data files You can then use the full LCMgui with the numerous possibilities documented in Chaps 6 amp 7 You can try setting everything up for possible future routine analyses and archival e When you click Run LCModel in Fig 6 4 you may or may not be asked to select a Basis Set Select any file e g HOME 1cmodel basis sets dummy basis LCModel will exit before it is used e LCModel will still only output the absolute value plot with the licensing message but the files will now also be archived as specified in Sec 7 3 3 You can set up and test the archival system and the rest of the configuration that you would use with a valid license 4 2 6 Starting Over LCMgui remembers what you have done so far You can make as many test
23. e g 24 25 26 27 You could incorporate most of these into LCMgui either with a Preprocessor Sec 7 4 5 or with your own my bin2raw Sec 7 4 4 10 1 CALIBRATION PHANTOMS 127 10 1 2 Scanner Calibration inter hardware Even if your Basis Spectra were scaled as in Sec 10 1 1 there will still be a second unknown calibration factor if they were acquired on another scanner Furthermore the properties of your scanner may change with time and you must periodically redetermine the factor scaling your Basis Set to data acquired on your scanner By using the simple method below you can compute the following calibration factor FCALIB REAL factor for calibration This will multiply your data being analyzed to correct for scaling differences between your Basis Set and your data Default FCALIB 1 0 This could also be done by multiplying TRAMP in the RAW file by this factor but FCALIB is more convenient and it is documented in your One Page Output Make a calibration phantom with 50 mM NAA in the solution in Sec 8 2 4 except that DSS and formate are unnecessary Don t forget NaN3 if you want to store the phantom at low temperature You will also have to bring the pH up to 7 2 e g with NaOH Analyze a spectrum from this phantom with LCModel using the following Control Parameters in addition to the usual HZPPPM NUNFIL DELTAT FIL etc which are usually automatically set by LCMgui NCALIB 1 CHCALI
24. following in Namelist NMALL HZPPPM NUNFIL DELTAT Same as in Sec 5 3 1 8 6 RUNNING MAKEBASIS 93 FILBAS FILPS AUTOSC AUTOPH CHARACTER 255 the name of the BASIS file to be output by MakeBasis If you are going to use LCMgui then follow the naming rules in Sec 7 3 2 CHARACTER 255 the name of the PostScript file to be output for plotting the real and imaginary parts of the Basis Spectra e g basis ps The spectra are properly scaled by your factor TRAMP VOLUME You can leave out FILPS only if you have suppressed the plots with PPMST and PPMEND below LOGICAL Auto Scaling TRUE to do Auto Scaling the formate stan dard must then be in all Basis Spectra See the definition of TRAMP above in Sec 8 6 3 for the requirements when AUTOSC F LOGICAL Auto Phasing TRUE to do Auto Phasing in this case the formate and DSS standards must then be in all Basis Spectra and you must also input AUTOSC T You will want to input values for the following unless you want the default values to be used IDBASI PGNORM PPMST PPMEND CHARACTER 80 an optional name for uniquely identifying this Basis Set This will be output in the BASIS file and by LCModel Default IDBASI Same as in Sec 5 3 6 Same as in Sec 8 5 2 except that the default is PPMST 4 2 PPMST and PPMEND only affect the plots in the PS file they do not affect the Basis Spectra in the BASIS file at all
25. increase PPMEND to be about 0 3 ppm inside the end of the spectral range However to fully include the main lipid peaks and some adjacent 104 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION baseline keep PPMEND lt 0 In future measurements be sure to increase the spectral bandwidth so that you can go down to 1 0 ppm At 2 9T you need a bandwidth greater than 1200 Hz you need at least 1800 Hz If you want to estimate E55 amp 153 the lipid components around 5 5 and 5 3 ppm you could input PPMST 8 0 Your residual water peak should not strongly overlap with 153 9 2 2 Additional Input 9 2 2 1 Dummy BASIS file LCMgui users can select any valid file name for the BASIS file in Fig 6 4 it is not used All basis spectra are automatically simulated inside LCModel For ex ample you could click Change BASIS in Fig 6 4 activate Show Hidden Files in the upper right of the BASIS file selector similar to Fig 6 1 Then select HOME 1cmode1 test output test basis This only needs to be set once 9 2 2 2 Ex Vivo spectra If your VOI is at room temperature input ROOMT T 9 2 2 3 Eddy current correction Normally you do not do eddy current correction ECC ECC assumes a pure water reference signal Large lipid signals invalidate this assumption and can cause artifacts LCModel automatcially switches off ECC for muscle and lipid spectra even if you have selected ECC You can force ECC by inputting FORECC T Wi
26. low 5 are output in boldface Some analyses can take much longer because of the search for optimal phase corrections The diagnostic info TWOREG 12 signals a very de tailed search 9 3 Lipid Spectra Liver Breast Bone etc The four SPTYPEs lipid 8 liver 11 breast 8 amp only cho 2 are similar and are specified together here They are summarized in Sec 9 1 It is essential that you read Sec 9 3 2 on output and carefully follow Sec 9 3 1 on input Additional instructions for only cho 2 to estimate choline only are in Sec 9 3 3 LCModel will not work with apodized spectra Sec 3 6 4 Spatial apodization with CSI is acceptable but not frequency spectral apodization 9 3 LIPID SPECTRA LIVER BREAST BONE ETC 107 9 3 1 Input LCMgui users can select any valid file name for the BASIS file in Fig 6 4 it is not used All metabolites are automatically simulated inside LCModel Section 9 2 2 1 gives a suggestion LCMgui users input the Control Parameters below using Fig 7 7 in Sec 7 3 4 You must always input the following SPTYPE 7 PPMST PPMEND 1 0 If your spectrum does not go down to 1 0 ppm i e you get the diagnostic FATAL INITIA 3 increase PPMEND to be about 0 3 ppm inside the end of the spectral range However to fully include the main lipid peaks and some adjacent baseline keep PPMEND lt 0 In future measurements be sure to increase the spectral bandwidth so that you can
27. the default will suppress creation of this file LCORAW 10 will make this file With LCMgui you only need to select CORAW in Fig 7 3 11 10 Analyzing a Time Series of Spectra You can analyze any series of spectra if you input them in the format of a CSI data set specified in Sec 5 2 3 1 LCModel will treat it as a CSI data set analyzing one spectrum after the other Often Philips series are automatically of this form with 1 row and N columns Check Fig 6 4 if necessary enter N in Total data Columns and right of through As with a CSI data set the analysis of spectrum j will be output in s11_1 3 ps If you want to use the Preview Data in Fig 6 4 to view a series of 64 spectra you can temporarily set the limits in the multi voxel panel for an 8 x 8 array for better viewing As with CSI data you can select a subset e g spectra 7 31 to be analyzed 11 11 Analyzing Data Left amp Right of the Water Signal If you want to analyze data both to the left and right of the residual water peak you can have LCModel insert a gap in the ppm axis being analyzed In this way you can cut out the huge meaningless residual water signal from the analysis without having to try to remove it with filters which inevitably cause distortions PPMGAP REAL 2 10 specifies that the J th gap in the analysis is from PPMGAP 1 J down to PPMGAP 2 J The gaps must not overlap and must be in strictly decreasing order Typical
28. when you are satisfied with the values below Restore Click on Restore to cancel any changes that you have made below eltat 5 120e 04 oecc t filbas SP sp lcmodel basis sets ge_steam_te30_95b basis filps SP sp icmodel temp 04d 00h 51m 51s 894pid ps filraw SP sp Icmodel temp O4d 00h 51m 51s 894pid metRAW key 123456789 unfil 2048 n owner My Laboratory My City ppmend 1 0 ppmst 3 85 S ddegp 2 title Exam 6089 8 ID FKESSLE1 08 03 98 16 26 steamcsi TE 30 TG R1 R2 98 6 30 2 0mL G01257 001 0000FKESSLE1 SDDEGZ 6 0 HZPPPM 63 87 Further restrictions for editing above 1 filraw filps etc contain temporary filenames assigned by LCMgui Do not change these 2 Each line must contain exactly one assignment e g you can enter nomit 1 chomit Glyc but not nomit 1 chomit Glyc 3 Do not assign a value to the same Control Parameter element twice This is easy to see since they are listed alphabetically Figure 7 7 You can view edit delete or add all Control Parameters here You do this by clicking on the Advanced Settings button in the Control Param eters window Fig 7 2 This produces the menu in Fig 7 6 You then click on View Edit Control Parameters in this menu to obtain the window in Fig 7 7 This shows the Control Parameters that will be input to LCModel You can edit or delete these or add new ones For example Fig 7 7 shows where I ha
29. 1 NAA FCALIB 1 0 NSIMUL 0 Let Ciem be the concentration output by LCModel and let Cirue be the true concen tration in the phantom e g 50 mM Then for your future analyses input FCALIB Cred Chan Repeat this calibration every few months at the beginning to see how rapidly it changes with time This FCALIB is not only restricted to this Basis Set but also to the bandwidth and acquisition time per scan of the calibration phantom acquisition LCModel can automatically adapt the Basis Set to changes in these but the scaling also changes So if for some reason you are using more than one bandwidth in time or frequency then you need an FCALIB for each 128 CHAPTER 10 ABSOLUTE METABOLITE CONCENTRATIONS 10 2 Water Scaling 10 2 1 Method With Water Scaling LCModel multiplies the data by fscale to scale the data consistently with the Basis Set i e so that the signal strength per proton resonance is the same in both This is the ratio of the normalized signal strength in the Basis Set to the normalized unsuppressed water signal in the data fscale Basisnorm Waternorm A normalized signal is the measured resonance area divided by the concentration of protons and by the factor lt 1 0 by which the measured signal is attenuated through relaxation etc Basisnorm Aredmet NIHMET x Concmet X ATTMET where Areamet is the observed resonance area of the selected singlet in the basis set by default the Cr CH3 N1
30. 142 CONTROL file LCMgui 60 CONTROL file 37 Control Parameters conventions 34 default 150 Hidden 151 LCMgui 63 Control Parameters Window 46 Control Defaults File LCMgui 65 COORD file LCMgui 60 COORD file 43 CORAW file 147 corrupted files 25 Cram r Rao lower bounds 15 132 cross correlation function referencing 136 CRUDE MODEL 155 CSF 102 CSI data sets 20 37 38 49 137 CSV file 42 139 CSV file LCMgui 60 DEEXT2 149 DEGPAP 95 DEGPPM 116 for MakeBasis 94 for PlotRaw 88 DEGZER 115 for MakeBasis 94 for PlotRaw 88 DELTAT 38 DESDSH 149 DESDT2 149 Detailed Output 149 DGPPMN 116 DGPPMX 116 Diagnostics Table 17 difference spectra 96 DOECC 115 DOREFS 136 DOWATR 108 DOWS 128 DSHPAT 122 dwell time 38 ECHOT INDEX for CONTROL file 140 for RAW file 35 eddy current correction 40 115 KECC program 86 LCMgui 48 EMCL 103 ERROR 155 error estimates SD 15 132 Execution Scripts LCMegui 71 false license 30 FATAL 155 FCALIB 127 Figures naming conventions 13 FILBAS for LCModel 37 for MakeBasis 93 FILCOO 43 FILCOR 147 FILCSV 42 File Selector LCMgui 45 filename filters 76 FILH20 115 FILPRI for LCModel 150 for PlotRaw 89 FILPS for LCModel 37 for MakeBasis 93 for PlotRaw 87 FILRAW for LCModel 37 for MakeBasis 94 for PlotRaw 87 FILTAB 42 filters data filenames 76 Final Analysis 115 FIXSHF 136 FMTBAS 93 FMTDAT 36
31. 3 4 Fixing the Shift The very first estimate of the starting shift is from the cross correlation function CCF described in Sec 11 3 5 The Preliminary Analysis then refines this If the spectrum is poor e g with the usual landmarks very weak the Preliminary Analysis may wander away and find a bad shift You can override the Preliminary Analysis and fix the shift to that from the initial CCF FIXSHF LOGICAL TRUE to fix the shift to be that from the CCF Default FIXSHF F except for FIXSHF T with SPTYPE muscle 5 since muscle is especially difficult to automatically reference The most drastic measure is if PPMSHF is also input with FIXSHF T then even the CCF is overridden and the shift is fixed at the input PPMSHF You must then be much more accurate with your input PPMSHF than when FIXSHF F 11 3 5 Details of Automated Referencing If you have to adjust the above settings from one analysis to the next then you might find a universal solution with the following settings LCModel determines the grid over which it will search for the referencing shift using the cross correlation function CCF with 9 functions described in 1 Appendix Two CCFs can be computed the first using the water peak even though it may be distorted due to water suppression and the second using the typical metabolite landmarks DOREFS LOGICAL 2 DOREFS JCCF F will exclude CCF JCCF If your residual water peaks are often strongly dist
32. As for FISHNI 1 9 FATAL The Hidden Control Parameter FRACT_POWER_SD must be positive SSRANG sum of squares range finds the range of the first sum in 1 Eq 4 correspond ing to a given range of confidence levels 11 1 ILLOGICAL Meaningless arguments in call to SSRANG 2 FATAL the computed upper limit in the range is smaller than the lower limit Hints This is only conceivable if you have modified the Hidden Control Parameter PRMNMX Do not input it STARTV starting values manages the elaborate Preliminary Analysis to get starting estimates for the Final Analysis 1 FATAL Both DOREFS 1 and DOREFS 2 are FALSE Thus no referencing can be done Hints This is a simple way to force an abort early in the run to get the Absolute Value Plot of the raw data on the One Page Output as in PLOT 5 in HOME 1cmodel doc figures pdf FATAL SHIFMN JREFS gt SHIFMX JREFS for JREFS 1 or 2 FATAL NREFPK JREFS is nonpositive or exceeds 10 for JREFS 1 or 2 Hints NREFPK 2 will be automatically decremented for each 6 function that is not within the Analysis Window If you drastically restricted the Analysis Window with PPMST or PPMEND then you may have to change PPMREF 2 and HZREF 2 so that at least one 6 function is in the Anal ysis Window Alternatively you can use PPMSHF which avoids using NREFPK warning The CCF with the first set of 6 functions specified by PPMREF 1 the water p
33. Control Parameters below or the CORAW button in Fig 7 3 will then use the phases and referencing shift from this LCModel analysis to phase and reference the data and output this into FILCOR in the format of a RAW file The simple text format of the RAW and CORAW files is specified in Sec 5 2 You can then take 11 10 ANALYZING A TIME SERIES OF SPECTRA 147 a weighted average over the series of CORAW files using your own script or program to produce an averaged RAW file for the final optimal analysis using the default Other button in Fig 6 3 Sec 7 4 4 The Namelist NMID has correctly set TRAMP VOLUME 1 0 since the data there have already been scaled Similarly any LCModel corrections called for by DOECC DOWS BRUKER SEQACQ or FCALIB are in FILCOR So you should be able to directly sum the CORAW files The only scaling necessary would be for statistical weighting if the original data files have different accuracies or scaling Use the Namelist NMID from a CORAW file for your averaged RAW file Note that the individual separate input files must be of sufficient quality to yield meaningful LCModel analyses Single scans may not be good enough FILCOR CHARACTER 255 corrected RAW file the name of the CORAW file to be output e g home user1 1cmodel saved pat_777_77 coraw LCMgui will auto matically output the file coraw in the directory that you specify in Fig 7 2 LCORAW INTEGER LCORAW 0
34. Control Parameters characteristic of your instrument If the list is too long to fit on the front page then by default a second page PLOT 2 in HOME 1cmode1 doc figures pdf with all the tables is printed It is best to use a printer that prints on both sides of a page otherwise Sec 5 3 6 shows how to easily suppress this second page but then the documentation will be incomplete The first three lines of this table in show the simple changes described in Sec 9 9 3 1 to get additional plots of the individual contributions of NAA and NAAG to the spectrum These are shown in PLOTs 3 amp 4 Chapter 3 Essential Guide This chapter briefly lists important guidelines and special cases It refers you to other chapters for details Print out HOME 1lcmodel doc figures pdf Remember figures located only in figures pdf are denoted by PLOT e g PLOT 6 Figures located in this manual are denoted by Figure or Fig e g Fig 6 2 3 1 Special Spectra LCModel attempts to automatically handle a wide variety of spectra with no special input from you especially if you use LCMgui However there are some special cases where you should input extra information to LCModel You input this information with Control Parameters Sec 5 1 2 With LCMgui these are simply entered in the Control Parameters window Fig 7 7 Without LCMgui they are input in Namelists Sec 5 2 1 Many Control Parameters have default values but you override these
35. File 5 3 1 5 3 2 5 3 3 5 3 4 5 3 9 5 3 6 5 3 7 5 3 8 Instrument and Acquisition Multi Voxel Data Sets Analysis Window Eddy Current Correction First Order Phase Correction PostScript Output Archiving the Results Phantoms os 4 2 6 a be ee hee a 6 Elementary Guide to LCMgui 6 1 6 2 6 3 6 4 6 5 Starting LCMgui Selecting the Data to be Analyzed 6 2 1 6 3 1 6 3 2 6 3 3 6 3 4 6 3 5 6 3 6 6 3 7 Where is the Output File Selector Window Do eddy current correction Do water scaling o a The Rest of the Control Parameters Window Reload Data Run LCModel 33 33 33 34 34 35 35 36 37 38 38 39 40 41 41 42 43 CONTENTS 6 6 Interactive Processing with LCMgui 51 6 7 Getting out of LOMBui e 51 LCMgui Reference Manual 53 TI HOM sur Structure uti A ee iaa 53 Mello EIMIDACIONS is ais tea ia bodied doy e eas 54 72 Tnstalline LCMepui i oos pat da Be ee a RA a es 54 7 2 1 Multi User Installations 4 54 22 SStarting ROMA coh ck lek Ae ala Oa A Oa ee Te s 55 20 First Test RUNS itean cites Ya eel wee as ro 55 7 2 4 Install Model Spectra o e e 55 2 0 Install License sisare to vce Che ee fe eS Sk Re A es 56 7 2 6 Installing Updates amp Upgrades 2 57
36. In the CONTROL file you could specify the first two elements of the KEY array as KEY 1 12345 KEY 2 23456 LOGICAL can be assigned the value TRUE or FALSE However you can also use T or t instead of TRUE and F or f instead of FALSE LCMgui and this manual use T and F e g BRUKER T 5 2 RAW File RAW is used to stress that the raw data should never be smoothed or windowed Smoothing destroys information and ruins the statistical tests in LCModel An example is provided by the RAW file with the test data in HOME 1cmodel test raw test RAW A RAW file consists of two Namelist headers preceding the time domain data itself as specified in the rest of this section 5 2 RAW FILE 35 5 2 1 Namelist SEQPAR The Namelist SEQPAR is optional but very useful If it is included its data will be compared with that in Namelist SEQPAR of the BASIS file to be used Diagnostics will then warn of inconsistencies between your in vivo data and your model spectra The RAW File begins with the Namelist SEQPAR SEQPAR ECHOT 20 HZPPPM 84 47 SEQ STEAM END Namelists are used throughout LCModel See your Fortran Manual for their many convenient features or use the Namelists in your LCModel package as guides Namelists are very easy to use The one critical pitfall the first column of Namelist input is always ignored so each line must start with one space The start of the Namelist is specified by a which mu
37. LCMgui exits and saves the changes that you have made to LCMgui settings such as the locations of your Basis and data files When the next LCMgui session is started these will then be offered as defaults Do Not exit to cancel the exit and continue with LCMgui Exit amp Forget LCModel exits but does not save any settings that you made during this session When the next LCMgui session is started you will be offered the same defaults that you were offered at the start of this session This can be helpful if you have made some big mistakes during this session or if you used some unusual settings that you do not want to have saved as defaults for the future New Profile allows you to save your settings to a new User Profile using the window in Fig 7 12 You should use different profiles for different types of data e g GE amp Philips or 1 5T amp 3T If the normal LCMgui exit buttons are not available then enter Control C or Control Z in the Linux Unix window from which you started LCMgui Chapter 7 LCMgui Reference Manual This chapter will allow you to fully exploit the flexibility of LUMgui It will also allow you to optimally install amp configure LCMgui for less experienced Normal Users You must be familiar with Chap 6 7 1 LCMgui Structure The main steps executed by LCMgui are 1 2 File Selector You select the file containing your binary scanner data to be analyzed with the window in Fig 6 1
38. LCModel calibration analysis with this new Basis Set The LCModel concentration that comes out of the calibration analysis should now be 1 0 Then GPC amp PCh are consistently scaled What happens if the numbers of protons contributing to the singlets are different Suppose CHBCAL GPC 9 protons and CHCALI 1 Cr 3 protons in the CH3 group Then you multiply the GPC con centration by 3 9 times the LCModel CONC that comes out of the calibration analysis With this corrected GPC concentration the calibration analysis should now yield 3 0 9 3 To calibrate NAAG with NAA use the above input except with CHBCAL NAAG CHCALI 1 NAA and no input for PPMST This is not as accurate because of the differing signals above 2 1 ppm However PPMST 2 1 often cuts too close to the base of the main peak and produces worse results So it is probably best to leave PPMST at its default of 3 85 or 4 0 Correct the NAAG CONC by multiplication as for GPC above LCMgui users can do the following With the File Selector in Fig 6 1 select an incorrect file type Click on Retry in Fig 6 2 Click on Other in Fig 6 3 Again select any reasonably small file with the File Selector in Fig 6 1 In the main Control Parameters window Fig 6 4 if necessary click on Change BASIS and select the Basis Set to be calibrated Also in the Control Parameters window click on Advanced Settings then on
39. NaOH or possibly with concentrated HCl the dilution will not be serious Otherwise you can simply correct each formate and metabolite concentration for the dilution At high fields if you had to reduce the phosphate concentrations then the reduced buffering capacity will require that you titrate especially carefully Without NaNz the solutions can go bad decomposition or microbiological growth within a few days Gln must be measured the same day as preparation even with NaN3 The NaN in the Standard Solvent allows you to store them longer some but not all for weeks at low temperature Let them return to room temperature before measurements To minimize oxidation GSH should also be measured quickly 8 3 ACQUIRING THE BASIS SPECTRA 83 Room temperature is often adequate for acquiring the model spectra below 3T With high resolution spectra e g 9 4 T 16 it is best to increase the phantom temper ature closer to 37 C At 3T and above this should be done at least with Cr amp PCr Obviously it is essential that the metabolite concentrations be accurately known The only exception is that the concentration of Sigma GPC is very uncertain because of the Cd complex Section 8 7 1 specifies how you can estimate the GPC concentration by analyzing its singlet as PCh whose concentration you accurately know 8 3 Acquiring the Basis Spectra 8 3 1 Quality The Basis Spectra should have a significantly higher S N and resolu
40. Page Output also has information on file locations SAVDIR Save Directory shows the directory where the output was archived SRCRAW Source of RAW file gives the full pathname of the data file that you chose with the File Selector SRCH20 Source of H20 file gives the full pathname of the unsuppressed water reference file if you chose one 6 6 Interactive Processing with LCMgui Normally you use LCMgui to start a series of analyses without waiting for one to end before starting the next However your analyses may go fast enough that you would want to run interactively viewing the LCModel results of one analysis on your terminal before starting the next Section 7 4 2 specifies how to do this If this has been implemented then you should wait for the results to be displayed on your screen To continue with LCMgui you must then exit the display program yourself 6 7 Getting out of LCMgui LCMgui usually has an Exit LCMgui or an Abort LCMgui button When you click this you are usually given the choice of four buttons Fig 6 6 52 CHAPTER 6 ELEMENTARY GUIDE TO LCMGUI O Exit Query TEE Exit a Save Exit normally saving your changes Do not exit Resume your session Exit 2 Forget Exit without saving your changes New Profile Create a new User Profile amp Exit Figure 6 6 Here you decide how to save the settings from this session Exit amp Save This is the normal exit
41. S N 18 183 simulating basis spectra 140 skipping voxels 139 smoothing data 23 34 S N 18 spectroscopic imaging 20 137 spreadsheets CSV file 42 60 139 SPTYPE 102 SRCH20 51 SRCRAW 51 Standard Solvent 82 starting over LCMgui 75 SUBBAS 122 TABLE file 42 TABLE file LCMgui 60 temporary files LCMgui 77 time series 147 TITLE 37 Toshiba data 26 36 62 145 TRAMP computing 125 in vitro 92 in vivo 36 tumor spectra 103 Unsuppressed Data 108 115 Updates installing in LCMgui 57 User Profiles 68 Varian data 25 27 62 VITRO 43 VOLUME in vitro 91 in vivo 36 warning 154 Water Scaling 128 LCMgui 48 WCONC 129 WDLINE for LCModel 121 for PlotRaw 89 WSMET 129 WSPPM 129 XLABEL 89 XLEFT 121 XRIGHT 121 XSTEP 122 184 XTRASH 95 YBOTT 121 YEAR4D 152 YLABEL 89 YTOP 121 INDEX
42. Scaling only the concentration ratios e g GABA NAA NAAG are meaningful unless the sequence voxel size and total gain are assumed constant over the study Even then the units of the macromolecule estimate if you use PPMEND lt 0 6 are different from those of the metabolites in the upper part of the table The signal around 3 ppm has a significant contribution from macromolecules as well as GABA This apparent GABA concentration is often called GABA 8 Because of the simplified models the SD values are underestimated more than usual Therefore only SD values below 5 are output in boldface 9 5 Internal Logical Units LCMgui users can skip Sec 9 5 Table 9 1 summarizes the files used by LCModel They are defined in more detail in other parts of this manual The pathnames are specified by Control Parameters beginning with FIL listed in the column headed Filename You must always input the Control Parameters for the names of all files that you will use If you forget this input then no Fortran OPEN will be executed These files have internal Fortran unit numbers given by Control Parameters beginning with the letter L in the Unit column of Table 9 1 The only allowed input values of these Control Parameters are in the column headed Alt i e to enable output with a positive number or disable output with zero of a file It is essential that two files do not have the same positive unit number The entr
43. Scaling be sure to input ATTH20 0 86 as explained in Sec 9 4 1 4 SDAT users can usually skip this paragraph The MEGA PRESS data set is stored as a CSI data set The total number of spectra say 40 in the data set is given in the Only for multi voxel panel in Fig 6 4 With SDAT files this is usually the Total data columns and nothing has to be done However with DICOM files this is usually Total data Slices and you must instead enter this number as Total data columns and change Total data Slices to 1 since LCModel only analyzes 1 CSI slice at a time Of course you must also set 40 in Selecting Columns 1 through 40 You can preview the data set by clicking the Preview Data button You can get a squarer preview by first changing the rows and columns in the panel in Fig 6 4 from 1 x 40 to 5 x 8 The analysis is not affected by this You can also do a separate analysis of the off spectra with a standard PRESS basis set You input either IAVERG 32 to sum only the even spectra i e when the first spectrum is on or IAVERG 31 to sum only the odd spectra i e when the first spectrum is off If in doubt preview the data You can also try both the wrong choice will show almost no NAA 9 4 MEGA PRESS FOR GABA 113 9 4 1 3 Analyses of off spectra You can also do a standard PRESS analysis of an off spectrum However the scaling of the ME
44. Script that will be used is shown in green right below Selected File You can select another Execution Script for this and future sessions by clicking on its name below Click on Use Selected File below when you are satisfied with the green file Selected File CC imake batch nice standard Figure 7 14 Here you can select the Execution Script that will be started by LCMgui 7 4 2 Interactive Processing Normally you use LCMgui to start LC Model analyses in background without waiting for one to end before starting the next However your analyses may go so quickly that you would want to run interactively viewing the LCModel results of one analysis on your terminal before starting the next If you can display PostScript files on your terminal as discussed in Sec 7 4 1 then Sec 7 4 6 shows how to change the command for printing your LCModel output to a display command The one limitation of this interactive mode is that you should wait for one analysis to end before starting the next to avoid multiple displays popping up 7 4 3 Execution Scripts As outlined in Sec 7 1 the final step of LCMeui is to start an Execution Script in back ground By default this script is HOME 1cmodel execution scripts standard You should not edit this script However you can copy it and use it as a template for your own customized scripts It contains suggestions on possible changes You must also put any Execution
45. Scripts that you make into the directory HOME 1cmodel execution scripts The currently active Execution Script is shown in parentheses next to Change Ex ecution Script in Fig 7 6 If you click Change Execution Script the window in Fig 7 14 shows you how to select another Execution Script for now and for the future 7 4 3 1 Batch Processing Normally you use LCMgui to start LCModel analyses in background without wait ing for one to end before starting the next However if your workstation is slower 72 CHAPTER 7 LCMGUI REFERENCE MANUAL processing analyses than you are in starting them you may want to avoid overloading it with many simultaneous jobs The Execution Script nice helps by reducing the priorities of the LCModel analyses but a much more effective way is to use the Execution Script make batch as follows e Select the Execution Script make batch in Fig 7 14 e Prepare and start your analyses as usual clicking Run LCModel one after another Exit LCMgui and enter HOME 1lcmodel bin run batch Only then will the analyses be started one after another one at a time e The start and finish times as well as any run time error messages appear on your screen All run time messages including harmless ones go into HOME 1cmode1 temp batch messages 7 4 4 Processing Non Standard Data Files LCMgui s first computation step Step 2 in Sec 7 1 is to produce LCModel input files from the data f
46. Signa LX last time If you have data of different types e g GE and Siemens then you should use a different User Profile Sec 7 3 7 for each This way the default settings for each data type are preserved and kept separate e Select a completely irrelevant file e g p wrong type 7 in Fig 6 1 to produce Fig 6 2 If you selected the wrong User Profile at the start of session then you can click the Abort LCMgui button and then restart and select the correct User Profile Otherwise click the yellow Retry button in Fig 6 2 A window like Fig 6 3 appears Here you click on the appropriate type of data The File Selector then returns giving you another chance to select your data file 6 3 Control Parameters Window After you click the OK button in the File Selector the second of the two most important windows the Control Parameters window Fig 6 4 appears LCMgui remembers most of the settings from the previous session Thus Normal Users often can simply click the yellow Run LCModel button at the bottom to start the LCModel analysis Yellow is generally used to show you the usual path to take 6 3 CONTROL PARAMETERS WINDOW 47 a Select your data type JAE Click on the button appropriate for your data Bruker Pid file if necessary converted by you to analog mode with Bruker s convdta gt ws Signa 5 x Raw P files or Spectrum G files GE LX Signa LK or higher Raw P files Mar
47. When you run the SAGE LCModel interface the window titled LCModel Analysis appears Be sure to make the following changes in every session SDDEGZ 999 0 only for CSI SDDEGZ 3 0 only for single voxel data SDDEGP 1 0 Start PPM 4 0 End PPM 0 2 Section 3 4 1 discusses cases where you might want to use other values for End PPM PPMEND There is reportedly a configuration file where you can set these values once for all future analyses Missing from the Concentration Table in the output are the important Concentration Sums NAA NAAG and probably Cr PCr and the sum of all choline containing compounds GPC PCh amp Cho Sec 2 1 1 You will have to compute these sums yourself It is important to use only these sums e g NAA NAAG and not the much less accurate individual estimates e g NAA and NAAG Similarly if PCr is estimated compute the concentration ratios to Cr PCr instead of the ratio to only Cr 7 2 6 Installing Updates amp Upgrades Updates and upgrades are used here interchangably Updates to any parts of the pack age will normally be incorporated into a complete new updated LCModel LCMgui package not patches If you also wish to save your old LCModel package e g for perfect compatibility in longitu dinal studies before updating you can enter cp a HOME 1lcmodel HOME 1cmodel old Then in the future after updating you can switch to the old version with mv HOME 1cmodel HOME 1c
48. You can convert other types of data for LCModel Until a few years ago there was no LCMgui and everyone converted their own data to the simple LCModel format specified in Chap 5 The analysis is then performed by a single Unix command of the form HOME 1cmodel bin lcmodel lt my control as specified in Chap 5 LCMgui also has an Other button that allows you to use your own conversion script within LCMgui see Sec 7 4 4 Chapter 4 Installation and Test Runs The complete LCModel LCMgui package can be downloaded from the WWW site on Page 2 The simple instructions for installing and uninstalling the package and making the test runs are also on this site You install as a normal user not root with no special privileges using the commands tar xf lcm tar install 1cmodel where is linux or sun or sgi or dec Please note If you are using a remote computer and get a message about not being able to open or connect to a display see the suggestions on http lcmodel ca lcm test x shtml Section 7 2 gives further options for installing LCMgui e g for a group of users 4 1 PostScript Output The summary of each LCModel analysis the One Page Output is in a PostScript file You may want to print this out However for routine analyses on a large scale you will instead probably want to display these files in color on the monitor and archive them for easier future access than piles of paper LCMgui facilitates
49. account for only differences in these increases from one metabolite to the next These differences are often negligible compared to the larger broadening of the lineshape due to field inhomogeneity etc 11 14 Detailed Output This output with up to 132 characters per line and typically 2000 lines is a full documen tation of the analysis I used this during testing and development You may be interested in some of its contents which are briefly summarized as follows a The Namelist LCMODL which contains the final values of all the Control Parameters as described in Sec 11 15 b A table documenting the Basis Spectra used in the Preliminary Analysis c Each iteration in the elaborate Preliminary Analysis for starting values for the refer encing shift and phase corrections for the data A table documenting the Basis Spectra used in the Final Analysis e Each iteration in the search for optimal regularizors in the Final Analysis After each final iteration are the values of the parameters the last set in the Detailed Output are generally the final values These values for 1000 Shift in ppm 1000 for ez in 1 Eq 4 are useful for checking the compatibility of the Basis Set with the data The shift parameters for the different Basis Spectra are listed in the same order as in the a 150 CHAPTER 11 FINE POINTS Basis Spectra table in d Any shift that is more than 4o e in 1 Eq 4 i e more than 6 standard dev
50. an NSHIFT gt 25 It will be reduced to 25 still a very large value FATAL You have input DGPPMN gt DGPPMX FATAL Same as SETUP 13 which should have already caught this info The Preliminary Analysis did not search the whole grid for the ref erencing shift as finely as it normally does Hints When the Hidden Control Parameter QUICK T this diagnostic is unavoidable Otherwise you can avoid this by increasing NSHIFT but this would increase the computation time and the default NSHIFT 8 is usu ally adequate despite this diagnostic If a disastrous referencing error like PLOT 17 in HOME 1cmodel doc figures pdf occurs Sec 11 3 can help FATAL All nonlinear least squares analyses in the Preliminary Analysis failed Hints As in RFALST 3 FATAL An essential nonlinear least squares analysis at the end of the Pre liminary Analysis failed Hints As in RFALSI 3 warning All nonlinear least squares analyses in the Preliminary Analysis failed to fully converge to optima Hints As in RFALSI 3 ERROR You have input the Hidden Control Parameter NSIDMX gt 20 It will be reduced to 20 FATAL You have input the Hidden Control Parameter NSIDMN nonpositive or exceeding NSIDMX in STARTV 17 ERROR You have input the Hidden Control Parameter INCSMX nonpositive or exceeding 20 It will be increased to 1 or decreased to 20 info The Preliminary Analysis has estimated a lineshape so narrow that Ng in 1 Eq 1 would be
51. an analysis of the test data supplied with your LCModel package At the top is the title Normally LCMgui or another Pre Formatting Program au tomatically supplies this by copying identification information from a header in the original data file 2 1 Concentration Table These are the results of main interest to you The most important message of this chapter is that before looking at the concentrations themselves you should always look first at the entries in the SD column These are the estimated standard de viations Cram r Rao lower bounds expressed in percent of the estimated concen trations These SD estimates are only lower bounds Sec 11 1 but they are still the most useful reliability indicators As usual you must multiply them by 2 0 to get rough 95 confidence intervals i e the range that would contain the true value about 95 of the time Thus e SD gt 50 indicates that the metabolite concentration could well range from zero to twice the estimated concentration Thus the metabolite is practically undetectable with this data Several metabolites are often undetectable in normal subjects e SD 20 indicates that only changes of about 40 can be detected with reliability e g the approximate doubling of Gln Glu in several pathologies A SD lt 20 has been used by many as a very rough criterion for estimates of acceptable reliability However it is only a subjective indication not a rigorous limit e SD
52. con versions of your data as you wish but if for some reason you want to start again from the very beginning i e from the test run you can clear LCMgui s memory by entering rm fr HOME 1cmodel profiles You can then start again as though you had just downloaded and unpacked the whole package by entering HOME 1cmode1 1cmgui Then click the Repeat Test Run button 4 3 Test Runs without LCMgui Until a few years ago there was no LCMgui You can still use LCModel directly without LCMgui It is easiest to do the usual installation with the two commands at the beginning of this chapter If the test runs are not done e g because X windows is not running everything that you need will still be installed mainly in HOME 1cmode1 bin You can do the test runs without LCMgui with the commands cd HOME 1cmode1 bin testrun 32 CHAPTER 4 INSTALLATION AND TEST RUNS The One Page Output is then in the PostScript file HOME 1cmodel test output test ps On Linux systems run time messages about Skipping namelist Seeking namelist are normal and harmless Chapter 5 specifies how to prepare and analyze your data either directly with LC Model or by using the Other button Sec 7 4 4 with LCMgui 4 4 Benchmark Timings To compare several computers you can execute the following command on each HOME 1cmodel bin time testrun The test run will be executed and then the time will be output on
53. confusion of vari able types such as omitting a single quote for CHARACTER variables or misspelling a variable name Linux messages about Skipping namelist are normal 12 1 2 Arithmetic Exceptions Overflows amp Zero Divides should never occur Check whether the format of your RAW data agrees with FMTDAT with the wrong format huge numbers might be read that overflow when squared Plot the data using the Pre view Data button in LCMgui Sec 7 4 1 If necessary rescale your RAW data e g with TRAMP so that the maximum absolute value is below 10 and above 1078 12 1 3 Other Aborts with no One Page Output Missing One Page Output Errors can occur during the production of the One Page Output and suppress it If the run ends without producing a PS file although it should have check for an error message in your Linux Unix window If no message can be found rerun getting the Detailed Output One of the Standard Diagnostics listed in the summary of diagnostics at the end of the Detailed Output will tell you why the One Page Output could not be produced 12 2 Standard LCModel Diagnostics These are in the Diagnostics Table in the One Page and Detailed Output The 5 types in increasing order of severity are info information of a non standard condition that might be of interest you but generally does not endanger the analysis warning something has happened that could affect the final results 12 2 S
54. data 16 With poorer data the baseline becomes flatter because there is no longer sufficient information to determine the baseline detail and the concentration estimates systematically get worse Methods always using flat baselines can yield more reproducible consistently wrong results with deceptively low coefficients of variation Flexible baselines are needed to compensate for the inevitable variability 132 11 2 RELAXATION CORRECTIONS 133 particularly of macromolecule and lipid signals The higher coefficient of variation using the more realistic LCModel should yield more realistic statistical tests 11 1 1 Averaging Concentrations over a Series of Analyses In computing an average concentration of a metabolite from a series of analyses you should include all values even those with higher SD You could do a weighted average over the series of analyses 31 C gt 001 10 wj 1 0 oj SD x Cj 100 with an estimated standard deviation a C 1 4 wj The C are constrained to be nonnegative their minimum allowed value is zero When a C 0 SD cannot be easily estimated 11 and 8D is arbitrarily output as 999 You might take an unweighted average in this case The C of a metabolite that has zeroes will probably have a large relative uncertainty and not be useful anyway 11 2 Relaxation Corrections The main message of this section is to use short echo times TE lt 30 ms and long repetition times TR gt
55. falsifies the model itself Your phantom data at room temperature will be inconsistent with simulated basis sets using data from 37 C If you are analyzing in vitro data then use VITRO LOGICAL TRUE if the spectrum is an in vitro spectrum this will override some statistical tests that could otherwise be falsified with such data Default VITRO F If necessary increase AQ the acquisition time for one scan so that the time data have decayed into the noise In addition your phantom spectrum must be dominated by at least one of the fol lowing NAA Cr or Cho Otherwise you will have to set additional parameters as illustrated in Sec 11 6 So it is much simpler if your phantoms always contain significant amounts of at least one of the above 3 landmark metabolites Even simpler is when you are doing standard calibrations either of the scanner Sec 10 1 2 or of the Basis Spectra Sec 8 7 1 Then VITRO T and several other Control Parameters are automatically set Chapter 6 Elementary Guide to LCMegui This chapter gives the basic steps for the Normal User in using LCMgui with refer ences to Chap 7 for details Chapter 4 gives the simple steps for downloading and installing LCMgui Chapter 7 is a detailed reference allowing you to make full use of the flexibility of LCMegui and to optimally configure LCMgui for the Normal User LCMgui remembers many of the settings from the previous run If an expert user has already m
56. file are at the top of the table 5 3 1 Instrument and Acquisition HZPPPM REAL Hz per ppm the field strength in terms of the proton resonance frequency in MHz i e HZPPPM 42 58Bo with Bo in Tesla You should input it with an accuracy of at least four significant figures Default HZPPPM 84 47 1 984 T so you will almost certainly have to input your own value NUNFIL INTEGER N unfilled the number of complex pairs of data points in one scan without zero filling Default NUNFIL 2048 DELTAT REAL dt the time in s between two consecutive real points or two consecutive imaginary points This is just the sample time or dwell time or the reciprocal of the bandwidth Default DELTAT 0 0005 The acquisition time for one scan DELTAT NUNFIL should be long enough so that the in vivo time domain data decay into the noise For brain spectra at 1 5 T this is typically 0 8 s or 1 0 s At higher fields it is usually shorter 5 3 2 Multi Voxel Data Sets One voxel s data is immediately concatenated to the preceding one s We use the DICOM standard that the voxels in the file start at the top left and are ordered as in three nested loops with the columns in the inner loop the rows in the middle loop and the slices in the outer loop i e the voxels are row after row then slice after slice Thus the second voxel s data in the file is said to come from Column 2 of Row 1 of Slice 1 ass
57. less than NSIDMN 5 Ng will be increased to NSIDMN Hints Be sure that AQ your acquisition time for one scan is not so short that the spacing between spectral points is too large to sample your peaks finely enough If this diagnostic occurs often then you should increase e g double AQ You must then do this for all your acquisitions all your data must have the same AQ If it occurs only seldom then you have only occa sional high resolution spectra and nothing needs to be done ERROR The Preliminary Analysis has estimated a lineshape so wide that an Ng in 1 Eq 1 would effectively exceed NSIDMX INCSMX 11 x 15 165 Grid Points This is being reduced to NSIDMX INCSMX in increments of 172 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS 22 23 24 25 26 INCSMX Hints Check for very poor data or spectral resolution If the data quality is acceptable then follow the Hints for STARTV 24 FATAL Your PPMST or PPMEND or one of the Control Parameters in INITIA 1 is so unreasonable that your Analysis Window extended by the finite convolution in 1 Eq 1 extends outside your entire spectrum Hints Check these Control Parameters If they are OK then you will have to reduce your Analysis Window to fit further within your spectrum or acquire your data with a larger bandwidth ERROR The Preliminary Analysis has estimated a lineshape so wide that the convolution in 1 Eq 1 would extend beyond the entire spec
58. metnum metdenom The illegal ratio part of the CHRATO is also at the end of the Detailed Output 4 FATAL In one of the CHRATO expect is negative or stddev not positive 5 FATAL You input the Hidden Control Parameter FCSUM non positive Do not input it 6 warning All concentrations in an analysis were zero A constraint is going to be skipped Hints There is probably a big problem with the data and a later FATAL diagnostic is likely only allows analysis of the demo test data unless you have a license 1 FATAL You do not have a license for this Sun SGI or DEC Compaq work station or your system date is wrong Hints Check that you are using the licensed workstation Check that your input of OWNER and KEY exactly match what was given to you with your license Correct your system date setting gt 1 FATAL With Linux there are numerous diagnostic numbers 11 meaning unreadable license is the most common Hints a If you have a license then make sure that it is called HOME 1cmodel license and not license dat or license bin b Send me the file HOME 1cmodel data for licensing if it was pro duced and the PS file produced Performs eddy current correction 1 FATAL Same as INITIA 1 2 FATAL Unreasonable values of NUNFIL PPMCEN or the Hidden Control Pa rameters PPM_TRUNCATE_MAX or PPM_TRUNCATE_MIN outputs and counts diagnostics 158 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS 1 2 ILLOGICAL M
59. minimum ppm value of the data You probably will not have to input values for the following GLOSCA SDPNTS LOGICAL global scaling This affects only multi voxel plots TRUE for all spectra in the plot to have the same vertical scaling FALSE for each spectrum to be scaled individually to fill the whole frame of its subplot Weak spectra are therefore expanded Default GLOSCA T REAL standard deviation in points LCModel analyzes the unsmoothed data However to more reliably locate the referencing peak in the Basis Spectrum MakeBasis temporarily convolutes the spectrum with a smooth ing Gaussian with a standard deviation of SDPNTS Grid Points in the fre 8 5 PLOTTING RAW FILES WITH PLOTRAW 89 quency domain where one Grid Point is 1 2 NUNFIL DELTAT Hz The default value for SDPNTS is 1 0 in MakeBasis and 0 0 no smoothing in Plo tRaw If you want to plot smoothed spectra e g to see what MakeBasis sees when it looks for the referencing peak then you can input a positive SDPNTS e You can customize the plots e g for slides or figures by inputting values for any of the Control Variables in Secs 9 9 3 2 9 9 3 4 except PTOUTP RHOUTP RGBLIN SUBBAS and RHVERS WDLINE in PlotRaw has dimensions WDLINE 6 3 WDLINE JCURVE JPLOT determines the curve thicknesses as in Sec 9 9 3 4 where JPLOT 1 for the plot of the time domain data JPLOT 2 for the plot of the frequency domain data and JPLOT
60. most likely to occur in spec tra with few well defined peaks as in tumors or outer CSI voxels Prior knowledge on the expected value of phase corrections can greatly improve analyses of difficult spec tra as well as speed up the analyses LCModel attempts to build up this information during a multi voxel analysis of CSI data For single voxel data LCMgui already inputs this prior knowledge that only small phase corrections are needed with GE Probe P files However other scanners usually each have their own narrow distribution of first order phase corrections Section 5 3 5 explains how to make good use of this 3 6 4 Smoothing Don t do it LCModel will not work with smoothed data Smoothing destroys infor mation and ruins the statistical tests in LCModel Other digital filters particularly excessive suppression of water or Baseline cannot avoid imposing assumptions and distorting the spectrum Do not zero fill Spatial apodization of CSI data is OK but not spectral apodization 24 CHAPTER 3 ESSENTIAL GUIDE 3 7 Vendor Specific Guidelines To be certain you can test the conversion of your data files with LCMgui and the reading of these by LCModel Chapter 4 shows how to do this even without a license 3 7 1 Bruker Normally you have the files fid acqp and method in one directory and LCMgui handles these automatically Sec 7 3 1 Do not switch on zero filling or exponential filtering Turn on Eddy Current Correcti
61. off spectra using IAVERG 31 or IAVERG 32 only half the spectra in both the suppressed and unsuppressed are used and you do not double ATTH20 The T1 relaxation times of water and the metabolites are similar and would therefore approximately cancel Sec 10 2 1 so corrections are often neglected In principle you still would have to correct the absolute concentrations for the significant T2 relaxation of the metabolites In view of all of these uncertain relaxation corrections it is best to consistently use the same sequence and corrections Then even if you skip some corrections you should hopefully still be able to see increases or decreases e g with pathologies even if your concentrations are in your institutional units instead of mM 9 4 1 5 Basis set Your basis set should contain GABA Gln Glu GSH NAA amp NAAG You can acquire phantom data not recommended or simulate them Sec 8 6 5 specifies how to do both 114 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION Table 9 1 Control Parameters for Filenames and Logical Units Unit Default Alt Needed when none LCONTR CONTROL Always FILBAS LBASIS BASIS Always FILRAW LRAW RAW Always FILH20 LH20 H20 DOECC T or DOWS T FILPS LPS PS LPS gt 0 FILCOO LCOORD COORD LCOORD gt 0 FILCOR LCORAW CORAW LCORAW gt 0 FILPRI LPRINT PRINT LPRINT gt 0 FILTAB LTABLE TABLE LTABLE gt 0 FILCSV LCSV CSV LCSV gt 0 9 4 2 Output Without Water
62. or NOBASI with these SPTYPE FATAL When using an SPTYPE in Sec 9 2 or 9 3 you must not use fwmini lt 0 Sec 11 7 to simulate spectra in a basis set since there is no input basis set FATAL When using SPTYPE lipid 8 breast 8 or liver 11 all concentrations in the Preliminary Analysis were estimated to be zero Hints Check for very poor data FATAL Invalid values for Hidden Control Parameter NPOWER FATAL You cannot simulate additional model spectra when SPTYPE muscle 5 lipid 8 breast 8 or liver 11 warning With short TE you should have PPMEND lt 0 6 to include all the macromolecule and lipid models such as MMO9 which is used to impose 164 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS important soft constraints on the other concentrations If your TE gt 100 ms you can ignore this diagnostic MYCONT reads the Control Parameters and does preliminary checks 1 10 11 12 13 14 15 16 17 18 19 20 21 22 FATAL An end of file was encountered before your Namelist LCMODL in the CONTROL file could be read completely No One Page Output will be pro duced Hints See Sec 12 1 1 FATAL An error was encountered in your Namelist LCMODL in the CONTROL file No One Page Output will be produced Hints See Sec 12 1 1 FATAL You have not input anything for FILPS although you set LPS gt 0 No One Page Output will be produced F
63. peak is outside the range of your spectrum Hint Check your input of PPMCEN If it is correct then the water peak at 4 65 ppm may really be outside the range of your spectrum In this case you could try inputting the undocumented TAREAW 1 which does Water Scaling without the water peak FATAL Same as AVERAG 3 AVERAG combines phased array spectra or spectra to be averaged 1 2 I o FATAL Same as INITIA 1 FATAL You input invalid values for the Hidden Control Parameter NBACK Do not input it FATAL Unable to find or read the file with your unsuppressed water refer ence data Should never happen with GE Probe P files unless you prevent water reference scans FATAL You have eliminated all channels of your phased array data using the parameters in Sec 5 3 2 or Sec 11 4 6 Do not eliminate any let LCModel statistically average all channels FATAL Water reference signal could not be integrated the phased array data cannot be combined FATAL Noise analysis of a channel failed Hints Never Happened FATAL Same as AVERAG 6 WARNING You have eliminated some channels of your phased array data or spectra from your Philips MEGA PRESS data set using the parameters in Sec 5 3 2 or Sec 11 4 6 It is usually better to let LCModel statistically average all channels or spectra FATAL Same as AVERAG 6 FATAL Same as AVERAG 6 12 2 STANDARD LCMODEL DIAGNOSTICS 157 BETAIN COMBIS CONCPR DEMO
64. referencing shift exceeds 0 07 ppm you may have to adjust SHIFM 2 to bracket this shift or input PPMSHF See the Index to find definitions of the Control Parameters SHIFMN etc 2 In pathologies where Glyc is high e g some tumors you may want to add Glyc to the analysis This is discussed in Sec 8 2 1 1 3 Referencing landmarks PPMREF 2 are 3 03 3 22 1 28 amp 0 90 ppm i e lipids replace NAA 4 Because of the unpredictable spectra the Preliminary Analysis uses all metabolites rather than the usual NUSE1 ones 5 NRATIO 13 i e a soft constraint on NAAG NAA is added 6 NAMREL GPC i e the ratios to total cholines rather than to Cr will be output The above settings are only initial defaults You can overwrite them by inputting your own Control Parameters For example if you input NAMREL Cr SPTYPE tumor you overrule Setting 6 above 9 2 Muscle Spectra 9 2 1 Standard Input LCModel will not work with apodized spectra Sec 3 6 4 Spatial apodization with CSI is acceptable but not frequency spectral apodization Normally you use water suppression LCMgui users input the Control Parameters below using Fig 7 7 in Sec 7 3 4 For muscle spectra it is essential that you input the following extra Control Parameters SPTYPE muscle 5 PPMEND 1 0 PPMST 3 8 If your spectrum does not go down to 1 0 ppm i e you get the diagnostic FATAL INITIA 3
65. referencing shift for the Final Analysis where the final concentration estimates are made LCModel runs faster if you have prior information on the approximate phase corrections More importantly this prior information can rescue spectra with weak NAA creatine amp choline landmarks such as in PLOTs 6 amp 13 in HOME 1cmodel doc figures pdf that might otherwise fail although these particular cases also worked without this prior information As specified below you impose this prior information as an expectation value and its standard deviation Recommendations for several types of scanners were given in Sections 5 3 4 amp 5 3 5 Section 9 7 1 suggests how you can experimentally estimate the next four values DEGZER REAL your expected value for the zero order phase correction dy using the definition of DEGZER in Sec 8 5 2 in the range 0 0 360 0 Default DEGZER 0 SDDEGZ REAL the standard deviation i e the uncertainty in your expectation value DEGZER in degrees A normal Gaussian prior probability distri bution is used so py can wander about three standard deviations from its 116 DEGPPM SDDEGP CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION expectation value DEGZER if the data demand this The four degrees of strength with which you can impose this prior information are in decreas ing order of strength SDDEGZ lt 1 fixes Hp to DEGZER with a standard deviation of 0 0 not 1 0 This is not reco
66. singlet at 0 0 ppm but still dilute enough so that its weak resonances in the Analysis Window are in the noise 8 2 MODEL METABOLITE SOLUTIONS Table 8 1 Concentrations and Allowed Names of Model Metabolites Metabolite L Alanine Aspartate Creatine Phosphocreatine y Aminobutyric Acid Glucose Glutamine Glutamate Glutathione GSH Glycerophosphocholine Phosphocholine myo Inositol L Lactate N Acetylaspartate N Acetylaspartylglutamate scyllo Inositol Taurine Acetone Acetate Betaine B hydroxybutyrate Carnitine Choline Citrate Cysteic Acid Ethanol Fucose Glucone Glucoronic Acid Glycine Guanidinoacetate Isoleucine Leucine Phenylalanine Phosphorylethanolamine Propylene Glycol Pyruvate Succinate Threonine Valine Allowed Conc Name s Ala Asp Cr Cre PCr GABA Gle Gln Glu GSH GPC PCh Ins mI Lac NAA NAAG Scyllo Tau mM Singlet ppm it protons Vendor 81 Order 05129 A9256 27890 A2129 49139 49419 49449 G8005 P0378 57570 71718 00920 153036 T0625 82 CHAPTER 8 MAKING THE BASIS SET 8 23 Phantom Use a spherical phantom of plastic or glass with about 300 mL volume smaller for animal scanners A simple way is to use normal spherical flasks with no flat bottom and cut or grind the neck off The phantom should be full and without air bubbles 8 2 4 Solutions MakeBasis can automatically phase and scale your spectra consistently with
67. the BASIS file does not match your data For the first runs you should nevertheless check whether the entry is correct STEAM or PRESS and TE should be visible in the filename If it is wrong you cannot edit the white field only green 48 CHAPTER 6 ELEMENTARY GUIDE TO LCMGUI 8 Control Paraneters 413 You can change any parameters below When you are satisfied click on Run LCModel at the bottom Non Expert users with single voxel data can often click on Run LCModel immediately me ATTE TOS HEE T1O7 S214 T80 See Aea ENZO 1927 ea Se TSS MEL TRENES Gn M OST BO RESEARCH CLINIC Analyzing spectrum from 4 0 H ppm down to 1 8 a ppm BASIS file SP sp Icmodel basis sets siemens_press_te1 35_99e basis _Change BASIS Do eddy current correction 1 Do water scaling Save Fe ypes to directory TT _Peconigure Only for Multi Voxel or Multi Channel data files Total data Columns Te Selecting Columns EN through me for Preview and then Analysis Total data Rows fie Selecting Rows a through fie for Preview and then Analysis Total data Slices cay Selecting Slice JE for Preview and then Analysis Advanced Settings Run LcModel Preview Data Reload Data Exit LCMgui Figure 6 4 The Control Parameters Window fields Instead you click on the green Change BASIS button and you get a File Selector for your BASIS file which you select just as in Sec 6 2 1 6 3 3 Do eddy current correct
68. the phase corrections 1 ERROR Your input DEGPPM is not between DGPPMN and DGPPMX It will be changed to the closest of these two Hint Correct your input of DEGPPM DGPPMN or DGPPMX PLINLS Partially linear least squares organizes the nonlinear least squares analyses There are many independent analyses and diagnostics here often only mean that LCModel is working hard but that the final results are not affected 1 ILLOGICAL Meaningless argument in call to PLINLS 2 FATAL Meaningless value for at least one of the following Hidden Control Parameters PMQST PMQSTL RSTPMX RCONVR RMQDEC RMQINC COSMIN RINCRS RPMQMN RSTPMN Hints Do not input these 3 warning Somehow a perfect fit has been achieved Hints Never Hap pened but if it did it would probably be due to inputting data that were all zero or equal to a Basis Spectrum 4 info Numerical instabilities have caused a normally nonnegative dot prod uct to become negative corrective action is taken Hints Working Hard 5 warning Somehow the magnitude of the step or the gradient of the objec tive function has become zero corrective action is taken Hints This can happen when the Marquardt parameter gets very large Working Hard 6 ERROR Somehow the denominator of a formula computing step size has become zero corrective action is taken Hints Never Happened 7 info Maximum iterations before complete convergence in one analysis Hints Working H
69. wrong model Smaller dips and humps in the Baseline are normal particularly compensating for item 4 around 2 0 2 3 ppm and below 1 8 ppm Weak examples of this are in PLOTs 1 amp 7 Similarly the Baseline in PLOT 18 handles the huge artifact dip fairly well The Baseline dip follows the data it does not sag down away from the data However the results still should be rejected the baseline is still too wild 3 6 CAUSES amp REMEDIES FOR FAILURES 23 3 6 Causes amp Remedies for Failures 3 6 1 Artifacts If the artifacts seem to be confined to an edge of the Analysis Window then the obvious remedy is cut them off by restricting PPMEND or PPMST as discussed in Sec 3 4 With CSI this is often done by inputting PPMEND 1 8 PLOT 16 in HOME 1cmode1 doc figures pdf Improved sequences might also eliminate the artifacts 3 6 2 Referencing The three major landmarks for initial referencing are at 3 22 ppm cholines 3 03 ppm Cr amp 2 01 ppm NAA Tumors are particularly difficult because Cr amp NAA are often both weak Referencing errors are usually easy to recognize and correct by inputting extra in formation Section 11 3 explains how to do this If the central frequency of your spectrum is more than about 0 1 ppm from the water resonance 4 65 ppm then referencing may fail and you may have to input PPMCEN as specified in Sec 11 3 1 3 6 3 Phasing Phasing errors are quite obvious and also rare They are
70. 0 ms At higher fields it is usually shorter Consistently use the same AQ for all data with a given sequence e Bandwidth should be at least 1000 Hz at 1 5T 2000 Hz at 3T and pro portional to these at other field strengths 3 4 ANALYSIS WINDOW 21 3 4 Analysis Window Two of the most important Control Parameters define the Analysis Window the window of frequency domain data that will be analyzed The left edge of the window is PPMST ppm 4 0 by default The right edge is PPMEND ppm normally 0 2 by default 3 4 1 PPMEND Normally you should leave PPMEND at its default of 0 2 However there are two important cases when this default should be changed 1 long echo times When LCMgui or LCModel detects TE gt 100 ms it resets the default to PPMEND 1 0 because the macromolecule signal around 0 9 ppm is usually too attenuated to be useful at such long TE You can override this default by inputting your own PPMEND 2 CSI data with lipid artifacts Strong lipid signals often occur in voxels near the edge of the slice This is no problem if the lipid signals are where they should be around 1 3 amp 0 9 ppm as in PLOT 6 in HOME 1cmodel doc figures pdf The problem is that some of these signals can originate from outside the voxel and can be significantly shifted or out of phase These artifacts distort the analysis in PLOT 15 With such artifacts there is no hope of estimating Lac Ala or anything else below 1 8 ppm PLO
71. 01 V Govindaraju K Young A A Maudsley Proton NMR chemical shifts and coupling constants fro brain metabolites NMR Biomed 13 129 153 2000 S A Smith T O Levante B H Meier R R Ernst Computer simulations in magnetic resonance An object oriented programming approach J Magn Reson 106a 75 105 1994 J Pfeuffer I Tk c S W Provencher R Gruetter Towards an in vivo neurochemical profile Quantification of 18 metabolites in short echo time H NMR spectra of the rat brain J Magn Reson 141 104 120 1999 T Michaelis K D Merboldt H Bruhn W Hanicke J Frahm Absolute concentrations of metabolites in the adult human brain in vivo Quan tification of localized proton MR spectra Radiology 187 219 227 1993 L Hofmann J Slotboom B Jung P Maloca C Boesch R Kreis Quantitative H magnetic resonance spectroscopy of human brain Influence of composition and parameterization of the basis set in linear combination model fitting Magn Reson Med 48 440 453 2002 BIBLIOGRAPHY 177 19 20 21 23 24 25 26 27 28 S W Provencher Low bias macroscopic analysis of polydispersity in Laser Light Scattering in Biochemistry S E Harding D B Sattelle amp V A Bloomfield eds Royal Society of Chemistry Cambridge 1992 pp 92 111 F G Woermann P A Bartlett M A McLean G J M Parker G J Barker J S Duncan Tailored single voxel short TE MR spect
72. 1 1 Metab1 CHUSE1 2 Metab2 CHUSE1 3 Metab3 3b More accurate and convenient is to use NCALIB 3 CHCALI 1 Metab1 CHCALI 2 Metab2 CHCALI 3 Metab3 and NSIMUL 0 Then NUSE1 CHUSE1 VITRO are automatically set for you 11 7 Simulating Basis Spectra By default LCModel simulates mainly lipid amp macromolecule Basis Spectra that are also included in the LCModel analysis You can change these with the following 11 7 SIMULATING BASIS SPECTRA 141 NSIMUL CHS IMU INTEGER is the number of Basis Spectra that you will simulate Default NSIMUL 13 CHARACTER 60 528 specifies the simulation parameters Defaults CHSIMU 1 Lipi3a 1 28 01 FWHM 15 lt 2 035 AMP 2 CHSIMU 2 Lip13b 1 28 01 FWHM 089 lt 09 035 AMP 2 CHSIMU 3 Lipi3c 1 30 01 FWHM 089 lt 09 035 AMP 2 CHSIMU 4 Lipi3d Q 1 26 01 FWHM 089 lt 09 035 AMP 2 CHSIMU 5 Lip09 89 02 FWHM 14 lt 19 035 AMP 3 CHSIMU 6 MMO9 91 02 FWHM 14 lt 17 015 AMP 3 CHSIMU 7 Lip20 2 04 005 FWHM 15 lt 2 025 AMP 1 33 2 25 FWHM 15 AMP 67 2 8 FWHM 2 AMP 87 CHSIMU 8 MM20 2 08 005 FWHM 15 lt 18 01 AMP 1 33 2 25 FWHM 2 AMP 33 1 95 FWHM 15 AMP 33 3 FWHM 2 AMP 4 CHSIMU 9 MM12 1 21 01 FWHM 15 lt 2 02 AMP 2 CHSIMU 10 MM14 1 43
73. 1 for a metabolite then you can use either If you introduce a new metabolite then use a Metabolite Name that is not in the table You must be consis tent even with case LAC will not be recognized as Lac Input without blanks i e gt Lac and not Lac Same as in Sec 8 5 2 This is only needed when AUTOPH F Same as in Sec 8 5 2 This is only needed when AUTOPH F You must have DEGPPM lt 200 but it should be much less shift the time domain data if necessary REAL the concentration of the metabolite in the model solution The units e g mM must be consistent with all other Basis Spectra REAL the concentration of the formate concentration standard in the model metabolite solution This is only needed when AUTOSC T Nor mally it is the same since the same Standard Solvent was used as solvent for all metabolites as specified in Sec 8 2 4 However it can vary when i a metabolite solution was diluted by titration ii combining Basis Spectra from different times or laboratories or iii when the phosphate Standard Solvent could not be used as with GPC CdClz or the Ca salt of PCh and a slightly different formate con centration was added The units e g mM must be consistent with all other Basis Spectra LOGICAL no shift TRUE to suppress referencing This can be useful with simulated basis spectra that are already referenced You still have to input sensible values for PPM
74. 11 4 1 Most other 2D amp 3D CSI data sets can be handled by LCMgui See Secs 3 7 amp 11 4 PLOT 14 in HOME 1cmodel doc figures pdf shows an LCMgui absolute value display of a full slice with 8x8 voxels A laser printer can usually give useful plots of up to 32x32 voxels With LCMgui you can then select and display a rectangular subset of the voxels in this slice and have LCModel analyze this subset or the whole slice in one multi voxel run An advantage of this is that LCModel first analyzes a central voxel of the subset and then works outwards using Bayesian learning to get starting estimates and soft constraints for the first order phase correction and the referencing shift from the preceding often better central voxels for the often poorer outer voxels This speeds up and improves the analyses LCModel outputs concentrations from all voxels in the subset to a file that can be imported into spreadsheet programs Sec 11 4 4 You should still be selective with your subset It is a waste of time to include outer rows or columns with only poor spectra Analyses of poor quality voxels can take several times longer than normal voxels 3 3 Acquisition Parameters e Apodization Never apodize the spectral data it distorts the data and ruins the statistical tests in LCModel See Sec 3 6 4 e AQ the acquisition time for one scan should be long enough so that the data always decay into the noise At 1 5T use AQ gt 80
75. 3 for the plot of the absolute value of the frequency domain data only JCURVE 3 5 6 are relevant in PlotRaw NAIM INTEGER 3 the number of numerical labels on the abscissa will be roughly NAIM JPLOT where JPLOT has the same meaning as for WDLINE JPLOT immediately above Default NAIM 8 18 18 XLABEL YLABEL CHARACTER 3 80 CHARACTER 2 3 80 label the abscissas and ordinates You can deduce their usage by outputting Namelist PLTRAW us ing LPRINT below LPRINT INTEGER You can output the Namelist PLTRAW to see all default values by inputting LPRINT 6 Default LPRINT 0 i e no file will be created FILPRI CHARACTER 255 the name of the file that will contain the output Namelist PLTRAW only needed if LPRINT gt 0 For CSI RAW files only NDSLIC NDCOLS NDROWS ISLICE IROWST IROWEN ICOLST ICOLEN have the same usage and defaults as with LCModel Sec 5 3 2 8 5 3 PlotRaw Diagnostics If the run stops before a PlotRaw diagnostic appears on your screen and no PS file is produced then check the suggestions in Sec 12 1 1 Normally a standard PlotRaw diagnostic of the form Aborting due to error Name n Check User s Manual appears on your screen If not it can always be found at the end of the PS file if some or all of the plots are not produced The following lists all the Name n possibly with suggested remedies if they are not obvious PLTRAW 1 You must have PPMST gt PPMEND 2 PAGEWD a
76. 4000 ms This avoids or at least reduces the complications below In practice the corrections below are often ignored anyway because of the uncertain ties in the relaxation times particularly in vivo and especially in pathologies You can still attempt to use uncorrected concentrations to establish norms provided that you keep TE and T R fixed If the relaxation times do not change much during a study then you may still get consistent measures of absolute concentrations in in stitutional units all deviating from mM by the same unknown factor These can then be useful at detecting increases or decreases in pathologies etc Concentration ratios are less sensitive than absolute concentrations to relaxation effects With acquired basis sets you would also have to include the in vitro relaxation For example you account for T2 relaxation by multiplying your LCModel concentrations by exp TE T2vitro _ ae r E 1 exp T E T 2vivo T2vivo T2vitro Simulations usually have effective in vitro T2 oo and T 1 0 so that only the in vivo relaxation must be accounted for Concentration ratios have the advantage that fra 134 CHAPTER 11 FINE POINTS the exponents contain only differences of relaxation times between the two metabo lites If you use Water Scaling Sec 10 2 you can probably ignore the TR correction below and in ATTH20 in Sec 10 2 2 since the two corrections approximately cancel Oth erwise y
77. 5 3 98 CHAPTER 8 MAKING THE BASIS SET 8 6 7 Output BASIS File As in the test runs in Chap 4 this output file can be input directly without change to LCModel and there is normally no need to read this subsection The BASIS file has the following general structure e Namelists SEQPAR amp BASIS1 only once at the beginning then for each Basis Spectrum e Namelist NMUSED used only to document your input to MakeBasis e Namelist BASIS used by LCModel e the real and imaginary parts of the Basis Spectrum obtained after phasing and zero filling ordered as real imaginary real imaginary The only new names in Namelist BASIS1 are BADELT REAL Basis t equals DELTAT input in Namelist NMALL in the IN file NDATAB INTEGER no of zero filled data points in Basis equals 2 NUNFIL input in Namelist NMALL in the IN file The only new name in Namelist BASIS is the referencing shift ISHIFT INTEGER the number of Grid Points that the Basis Spectrum will be shifted for correct referencing A positive ISHIFT shifts the spectrum toward higher ppm values 8 7 Calibrating Basis Spectra 8 7 1 Calibrating GPC PCh amp NAA NAAG Auto Scaling cannot detect or fix errors in CONC your concentration values Such errors are directly propagated to the LCModel concentrations However they are at least reproducible so that you can still see changes or trends A worse situation is when two very similar metabolite
78. 7 Lip13 in CHRATO 1 becomes Lip13a Lip13b and MMO9 in CHRATO 3 becomes MMO9 totCr in CHRATO 7 will sum all that are present of Cr Cre amp PCr totCho will sum all that are present of Cho GPC amp PCh totNAA will sum all that are present of NAA amp NAAG Big3 in CHRATO 8 is a special symbol for totNAA totCr totCho totCho totCho To make Big3 more invariant totCho is added three times because it is often present when the totNAA amp Cr are weak as in tumors totCho also has three times as many protons per molecule contributing to its main peak expect the expectation value of the ratio stddev the standard deviation of expect metwt is optional It will be included in the initial estimate of the weighting of the nonlinear ratio constraint It should include metabolites that could be mistaken for metdenom e The tokens WT totCho totCr totNAA Big3 must be exactly as above with each CHRATO on one line There must be no space on either side of The necessity of the Concentration Ratio Priors for the lipid amp MM components is ex plained in Sec 11 7 1 The prior for CrCH2 totCr reflects typical attenuations of the Cr CH signal by relaxation and water suppression By default CHRATO 13 is only used when SPTYPE tumor or csf The priors for the other ratios are extremely weak with stddev typically 100 of expect The Asp GABA Glc Scyllo amp Tau signals are often so weak that it is tempting t
79. 7 3 Basic Settings and Usage o e a 58 7 3 1 Vendor Specific Guidelines 2 eee 58 ES BASIS lO A A a Be a A AE 60 7 3 3 Archiving the LCModel Results 60 7 3 4 View Edit Control Parameters o o 63 7 3 5 Saving Your Control Parameters 65 7 3 6 Changing your Default Control Parameters 67 Tose User Profiles e pasior ete ate hee atta a ak 68 7 4 Further Useful Settings 0 0 0 02 02 20 2000 4 69 7 4 1 Previewing Data 00002 eee 69 7 4 2 Interactive Processing e e e 71 7 43 Execution Scripts e 71 7 4 4 Processing Non Standard Data Files 72 7 4 5 Optional Preprocessor 000 o 73 7 4 6 Changing the License amp the Print Command 74 Lor ainesPoimts a a ts Bos ee ee ag 75 7 5 1 gui defaults Resource File 75 7 5 2 Cleaning up the File System 04 77 CONTENTS 8 Making the Basis Set 8 1 8 2 8 3 8 4 8 5 8 6 8 7 8 8 Compatibility Requirements 0 0000 eee eee Model Metabolite Solutions o e o o 8 2 1 Choice of Model Metabolites 8 2 2 Reference Markers o e e 82 3 Phant m mada eee da a 8 24 Solitons te tala a a a Se dt el a Acquiring the Basis Spectra e o Sisal QUAY cto Ea dee sy A
80. 8 MAKING THE BASIS SET range PPMOFF 1 gt PPMOFF 2 is zero So you set this range where there is only Baseline in the Basis Spectrum It should also be in the Analysis Window For example for NAA you might use the range 3 8 3 2 or 1 6 1 0 Default PPMOFF 999 999 i e meaningless values so that PPMOFF is not used You must explicitly input all of the above Control Parameters in each Namelist NMEACH except that e You must omit DEGZER and DEGPPM when AUTOPH T e You may omit PPMPK XTRASH DEGPAP NOSHIF and PPMOFF to leave them at their defaults e You may omit CONCSC when AUTOSC F 8 6 5 MEGA PRESS Because of the limited accuracy of MEGA PRESS and probably most other dif ference spectra you can use a shifted NAA to approximate NAAG with XTRASH Sec 8 6 4 3 If you simulate the basis set you can skip the rest of this section You must make a basis set from the difference spectra but the usual landmark singlets of formate amp DSS are practically absent So you use the unedited off spectra for phasing referencing and scaling 1 Make a basis set from your unedited off spectra Use AUTOSC T AUTOPH T 2 Make your final basis set from your difference spectra using the same Con trol Parameters as in 1 except 2a AUTOSC F AUTOPH F NOSHIF T in each namelist NMEACH and 2b Use the values from the basis file from 1 for DEGZER DEGPPM in each namelist NMEACH 3 Edit
81. A PRESS FOR GABA 111 9 3 3 1 Referencing There is one prerequisite water suppression must be balanced as follows It must be strong enough so that the baseline around 3 8 ppm does not climb or fall towards the residual water signal It must be weak enough so that the residual water peak is not significantly shifted from the original i e the maximum must still be within about 0 05 ppm of 4 65 ppm This is necessary because the water peak is used for preliminary referencing Cho cannot be used since it can be undetectable e g in healthy tissue However if the water suppression is too strong and the water peak is too distorted you could try referencing with Cho by inputting DOREFS 1 F DOREFS 2 T SHIFMN 2 shift 0 2 SHIFMX 2 shift 0 2 where shift is the expected value of the referencing shift often zero Extend the above two by 0 2 if the possible range Sec 11 3 5 is greater than 0 2 ppm but keep them as small as your prior knowledge allows A more direct way is to specify the referencing shift with PPMSHF Sec 11 3 3 9 3 3 2 Rejection criterion Even when Cho is absent you will often get a small peak fitting noise This is also normal with weak metabolites in brain spectra The difference is that with this empirical fit to a single peak the SD value is underestimated Sec 9 3 2 and cannot be used as a rejection criterion Sec 2 1 Instead you might consider a signal noise S N lt 2 to indicat
82. APP although they will not be used Default NOSHIF F 8 6 RUNNING MAKEBASIS 95 PPMPK PPMAPP XTRASH DEGPAP PPMOFF REAL ppm peak the chemical shift in ppm of the resonance that will be used for referencing Default PPMPK PPMPHA i e usually 0 0 for DSS REAL 2 ppm apparent You may want to first plot the absolute value of your spectrum with PlotRaw Sec 8 5 or with LCMgui s Preview Data Sec 7 4 1 From this plot you must specify with PPMAPP the interval on this apparent unreferenced ppm axis that contains your referencing peak The program MakeBasis will find the maximum on this interval and compute the shift that will translate this maximum to the correct PPMPK Note that PPMPK PPMSCA PPMPHA PPMST and PPMEND are all true ppm values after any necessary referencing only PPMAPP refers to apparent unreferenced values PPMAPP 1 is the high ppm end of the interval PPMAPP 2 is the low ppm end of the interval If DSS is the marker for the metabolite then PPMAPP 0 1 0 4 is typically sufficient to bracket the peak REAL extra shift is used to shift a complete spectrum along the ppm axis to simulate another spectrum XTRASH new doig in ppm Note that PPMSCA PPMPHA PPMAPP etc are not affected by XTRASH scaling phasing and referencing are correctly done before the extra shift although the plots in the PS file will show your formate and DSS peaks sh
83. ATAL You have not input anything for FILCOO although you set LCOORD gt 0 FATAL You have not input anything for FILTAB although you set LTABLE gt 0 FATAL You have not input anything for FILRAW or you input LRAW lt 0 FATAL You have not input anything for FILH20 or you input LH20 lt 0 although DOWS T or DOECC T or IAVERG 1 or 4 or UNSUPR T FATAL Your input for SPTYPE contains an invalid string Hints Correct your input of SPTYPE or do not input it FATAL You input one of the following non positive NDCOLS NDROWS NDSLIC ICOLST ICOLEN IROWST IROWEN ISLICE Hints 1 You must explicitly set these in LCMgui when LCMgui cannot determine these values e g from CSI data in Siemens Vision or ima files 2 This error can also be caused by incorrectly transferring files see Sec 3 7 5 FATAL Your input violates one of the following requirements ISLICE lt NDSLIC ICOLEN lt NDCOLS IROWEN lt NDROWS FATAL You have not input anything for FILCOR although you set LCORAW gt 0 FATAL NUNFIL lt 64 an unreasonably small value FATAL You must have 0 lt IAVERG lt 4 or IAVERG 31 or IAVERG 32 FATAL You must have 1 lt IROWSK J lt NDROWS and 1 lt ICOLSK J lt NDCOLS for 1 lt J lt NVOXSK FATAL You have not input anything for FILCSV although you set LCSV gt 0 Not used warning With SPTYPE muscle 5 lipid 8 breast 8 or liver 11 you shou
84. Append buttons below will append elements most that will change with each spectrum yielding unique directories You can also enter fixed elements in the green field below and click on Append Entry This can also be a single to separate subdirectory names The yellow field shows the Save Directory that you have constructed so far Click on Start Over to clear the yellow field and start constructing again So you can experiment and undo any errors by clicking Start Over If the yellow field ends with e g name then your files will be archived in name table name ps etc instead of name table name ps Click on Cancel to cancel your changes here and keep the old configuration Click on Set Configuration below when you are satisfied with the configuration in the yellow field This will then also be used for future default Save Directories Click on Restore Default below to restore the original default Save Directory configuration for your data type Bruker _Append Default Save Directory SP sp lemodel saved _Append Original Data Directory tmp lcm binary data bruker r0107 Enter a fixed element ETC Append Enty star Over Constructed Save Directory fnenbinary etalraveri iOvipdetaomodel O Cancel Set Configuration Restore Default Figure 7 4 The window for permanently reconfiguring the structure of Bruker Archive Directories 7 3 3 3 Reconfiguring the Archive Directory Tree You ca
85. Basis Set and your data will always satisfy the compatibility requirements in Sec 8 1 then you do not need to read the rest of this chapter Before acquiring model spectra check with me to see if a basis set compatible with your data is already available We have a large collection freely available for use with LCModel Your in vivo spectra are analyzed as a linear combination of these Basis Spectra Your efforts in getting a good Basis Set are well invested since this will be the basis of the analyses of probably thousands of spectra This is not a nice little project for a student It requires great care and expertise An alternative is simulating the basis spectra 12 15 8 1 Compatibility Requirements The following conditions in your in vivo data must be the same as those in the Basis Set e Bo field strength only to within a few e Localization sequence e g STEAM or PRESS e TE to within a few ms e TM with STEAM to within about 5 10 ms The following do not have to match TR the spectral bandwidth or the acquisition time per scan 78 8 2 MODEL METABOLITE SOLUTIONS 79 8 2 Model Metabolite Solutions 8 2 1 Choice of Model Metabolites Table 8 1 shows most of the metabolites that have been used with LCModel At low fields up to 2 35 T and moderate TE up to 80 ms all metabolites in the upper part of the table are recommended Note that you do not buy Glyc or any other metabolite with only a singlet in
86. CModel will start looking for 1 SDDEGP may of course further restrict this range If DEGPPM above is non zero input DGPPMN DEGPPM 30 0 DGPPMX DEGPPM 30 0 SDDEGP 30 0 1 5 Increase 30 0 above if the possible range Sec 9 7 is greater than 30 0 Defaults DGPPMN 30 DGPPMX 30 9 7 1 Estimating Appropriate SDDEGZ SDDEGP DEGZER amp DEGPPM The prior phasing information in SDDEG and DEG can speed up and improve your analyses You can estimate appropriate values for these by doing a series of LCModel analy ses on a large number of your best in vivo spectra Do not use any prior phasing information Compute the means and standard deviations of the LCModel phase corrections 3rd line in the MISCELLANEOUS OUTPUT table and use these means amp SDs for DEGZER DEGPPM and SDDEGZ SDDEGP The SDDEG will tend to be too 9 8 SPECIFYING THE BASIS SPECTRA FOR THE ANALYSIS 117 large because of the extra scatter due to the noise in your data that is the reason to use your best spectra Always use SDDEGZ gt 2 and SDDEGP gt 1 If LCModel strongly disagrees with your constraints in future analyses then you will get warning FINOUT 10 or FINOUT 11 Follow the suggestions for these diagnostics in Sec 12 2 9 8 Specifying the Basis Spectra for the Analysis 9 8 1 Omitting Basis Spectra from the Analysis Normally all Basis Spectra in the BASIS file are used by LCModel If some of these are very unlikely t
87. GA PRESS and off concentrations will be different since the basis sets are different Therefore to take even a concentration ratio of one metabolite from a MEGA PRESS spectrum to one from an off spectrum e g GABA Cr you must do Water Scaling Sec 9 4 1 4 in both analyses 9 4 1 4 Water Scaling For this you need in addition an unsuppressed water reference spectrum acquired from the same voxel immediately before or after the water suppressed spectrum It must be an off spectrum not a difference spectrum You input DOWS T With LCMgui you use Sec 6 3 4 Except for Philips you select the difference water suppressed spectrum to be analyzed and then as water reference the off unsuppressed spectrum With Philips you select the suppressed and unsuppressed SDAT files In all cases the 2 files must have identical format With Philips files they must have the same number of spectra The unsuppressed water relaxation attenuation is automatically accounted for by mega press 3 internally with the factor Secs 10 2 2 amp 11 2 ATTH20 exp TE T2 exp 68 80 0 43 For TE 80 ms you could input ATTH20 0 37 With Philips files and IAVERG 3 you must double this and input ATTH20 0 86 because two spectra on plus off are used to compute the water suppressed difference spectra whereas full water peaks are present in all the unsuppressed water spectra With a standard PRESS analysis of the
88. HMET is the number of equivalent protons contributing to the selected singlet Concmet is the model metabolite concentration in the basis set and ATTMET the attenuation factor of the singlet used in the in vitro basis set Similarly Waternorm Aredwater 2 x WCONC x ATTH20 As specified below N1HMET and Concmet are known the two Area terms are computed by LCModel from the data but ATTH20 and WCONC must be guessed 10 2 2 Usage With LCMgui You activate Water Scaling by activating the Do water scaling but ton in Fig 7 2 which sets DOWS TRUE DOWS LOGICAL Do Water Scaling DOWS T to activate Water Scaling You definitely know whether or not Water Scaling has been done by the presence or absence of the Doing Water Scaling message in the DIAGNOSTICS table in the LCModel output Default DOWS F Note that DOWS is automatically switched off overriding any input of DOWS if any of the following conditions exists No unsuppressed water reference is available NCALIB gt 0 or BASCAL TRUE indicating an internal basis calibration run FCALIB 1 0 indicating that the Calibration Phantom method of Sec 10 1 is being used WCONC and ATTH20 below are difficult to guess correctly This means that all of your concentrations will still be off by an unknown factor even if you also use Sec 11 2 to correct for metabolite relaxation You therefore obtain concentrations in unknown institutional units al
89. L All the curves to be plotted have exactly one single value Hints Never Happened but if it did this single value would probably be zero indicating gross input errors 9 FATAL Same as ONEPAG 8 10 ERROR All of the Residuals to be plotted have exactly the same value Hints Never Happened but if it did the value would probably be zero and the hints for PLINLS 3 or ONEPAG 8 would apply 11 FATAL Same as ONEPAG 8 12 ERROR There is not enough vertical space for even the Concentration Table Hints Check the Control Parameters in Secs 9 9 3 2 amp 9 9 3 3 if they are reasonable then input IPAGE2 1 OPENOU Opens and checks output files giving them unique names with multi voxel data sets 1 FATAL One of your output pathnames would be more than 255 characters long This probably happened inserting a unique identifier for multi voxel data Hints Shorten the pathname 2 FATAL Error trying to open FILPRI Hints Check permissions 3 FATAL Error trying to open FILCOO Hints Check permissions 4 FATAL Error trying to open FILTAB Hints Check permissions 5 FATAL Error trying to open FILCOR Hints Check permissions PARSIM Parses the string in CHSIMU JSIMU Except for error number O below the error number is JSIMU 100 x TERR which are listed below For example if you get FATAL PARSIM 709 then your CHSIMU 9 caused the diagnostic labeled 7nn below All PARSIM errors are FATAL Hints See Sec 11 7 for the symb
90. L There are more than 65536 voxels in your RAW file 3 FATAL Error trying to open FILRAW Hints Check for non existent file or permissions 4 FATAL Error trying to read Namelist NMID in FILRAW Hints See Sec 12 1 1 5 FATAL Error trying to read time data in FILRAW Hints In FILRAW check that the time data correspond to FMTDAT Does the file size correspond to your input NUNFIL or with multi voxel data with NDSLIC NDROWS NDCOLS Bibliography 1 2 S W Provencher Estimation of metabolite concentrations from localized in vivo proton NMR spectra Magn Reson Med 30 672 679 1993 W Hanicke T Michaelis K D Merboldt J Frahm On the use of a fully automatic data analysis method for in vivo MRS Metabo lite concentrations and relaxation times from proton spectra of human brain Proceedings of the Society of Magnetic Resonance in Medicine Twelfth Annual Meeting Society of Magnetic Resonance Berkeley 1993 p 977 U Seeger U Klose I Mader W Grodd T Nagele Parameterized evaluation of macromolecules and lipids in proton MR spec troscopy of brain diseases Magn Reson Med 49 19 28 2003 S Stockler U Holzbach F Hanefeld I Marquardt G Helms M Requart W Hanicke J Frahm Creatine deficiency in the brain A new treatable inborn error of metabolism Pediatr Res 36 409 413 1994 D P Auer C G ssl T Schirmer M Czisch Improved analysis of H MR spectra in the presence of
91. LCModel amp LCMgui User s Manual Stephen Provencher November 8 2015 1LCModel Version 6 3 1L LCModel LCMgui and their documentation are copyrighted 1992 2015 by Stephen Provencher All rights are reserved The latest version of the complete LCModel LCMgui package and the documen tation can be downloaded from the following site http lcmodel ca DISCLAIMER Please also read the License Agreement for LCModel LCMgui which is on http lcmodel ca 1cm license shtml The sole task of LCModel is quantitation of proton MRS data input in the standard LCModel format To help users most of the LCModel manual now covers areas far beyond this task For example e Information on data acquisition conversion and scaling for the numerous data and hardware types is based on information kindly provided by others It may not be up to date accurate or complete Consult the manufacturer of your hardware e Much additional information on possible applications is based on research still in progress and subject to error and revision The LCModel manual LCModel amp LCMgui User s Manual is available free of charge on the understanding that neither the author nor Stephen Provencher Incorporated nor any contributors of information to the LCModel manual shall be held liable for possible errors or omissions in the LCModel manual or the consequences thereof Contents 1 Preface amp Overview 11 1 1 What You Must Read 2 2 ee
92. LCModel LCMgui provides help with two different strategies Calibration Phantoms Data from a phantom of say NAA with a known con centration is analyzed periodically The Scale Factor is simply the factor necessary to make the LCModel NAA concentration come out correctly Sec 10 1 Generally can only be used with Siemens Vision 17 and 1 5 T GE data 20 but not phased array data An absolute method Still requires accurate absolute scaling of each spectrum for voxel size and changes in total gain due to coil loading etc LCMgui only does this for Siemens Vision amp GE data and even here on the assumption that methods in Sec 10 1 1 are exact 124 10 1 CALIBRATION PHANTOMS 125 Water Scaling The resonance area of the unsuppressed reference water signal to gether with the assumed water concentration is used 21 22 Sec 10 2 A ratio method Several sources of error such as voxel size and hard ware changes are automatically eliminated Relaxation effects espe cially T1 effects are reduced Very convenient just click the Do water scaling button but also check Control Parameter settings in Sec 10 2 2 No calibrations with phantoms Requires large uncertain corrections for water concentration and re laxation possible instrumental attenuation of the unsuppressed water signal etc Sec 10 2 2 e Automatically done by LCMgui with GE single voxel phased array data e For data other than
93. Lac are very strongly correlated If you want to also output the total Thr Lac concentration then with the above defaults you input NCOMBI 17 9 9 ONE PAGE OUTPUT 119 CHCOMB 17 Thr Lac Unlike NAA and NAAG the main peaks of the Glyc and Ins have different effective numbers of protons 17 which means that the Concentration Sum could have up to about a 15 error if Glyc is mistaken as Ins or vice versa So for consistency you normally do not have Glyc in the Basis Set Sec 8 2 1 1 it is poorly resolved from Ins anyway The same goes even more so for GABA and Glu Gln The GABA signal is much weaker So you should not consider the sum GABA Glu Gln it can fluctuate strongly because of the unreliable GABA concentration estimate At higher fields 3T and above and resolution Glu amp Gln are well resolved and the sum Glu Gln is unnecessary but it does no harm If the SD for the sum Glu Gln is lower than that of both Glu and Gln then Glu and Gln are strongly enough correlated that the sum Glu Gln is useful 9 9 1 2 Concentration Ratios You can output the concentration ratio relative to any desired metabolite as follows NAMREL CHARACTER 56 name for relative concentration the Metabolite Name corresponding to the concentration in the denominator of the ratio It can also be one of the Concentration Sums in CHCOMB Default NAMREL Cr CONREL REAL The third column in the Concentration Table is actually t
94. MCEN 134 PPMEND for LCModel 21 39 for MakeBasis 93 for PlotRaw 88 PPMGAP 147 PPMOFF 95 PPMPHA 94 PPMPK 95 PPMREF 136 PPMSCA 93 PPMSHF 135 PPMST for LCModel 21 39 for MakeBasis 93 for PlotRaw 88 Pre Formatting Program 38 Preliminary Analysis 115 140 Preprocessor LCMgui 73 Preview Data LCMgui 49 69 PRINT file LCMgui 60 PRINT file 150 printed output LCMgui 29 74 priors concentration ratios 143 relaxation amp shift 148 protecting configuration LCMgui 75 PS file LCMgui 60 PS file 34 37 PTLABL 121 PTOUTP 121 PTTITL 121 INDEX RATIPM 119 RAW file LCMgui 60 RAW file 34 rectangular subset multi voxel 39 referencing shift 136 correction 148 149 sign convention 98 regularization 142 relaxation corrections 133 relaxation priors 148 reproducibility 132 Residuals 17 22 RFWHM 148 RGB color specifications 120 RGBBOL 119 RGBERR 120 RGBLIN 122 RGBRAT 120 RHLABL 121 RHOUTP 121 RHTITL 121 RHVERS 121 rodents 145 ROOMT 108 SAGE LCModel interface 57 SAVDIR 51 Scale Factor absolute concentrations 124 MakeBasis 91 SD 15 132 SDDEGP 116 SDDEGZ 115 SDPNTS 88 94 SEQ 35 SEQACQ 36 SEQPAR Namelist MakeBasis IN file 92 RAW file 35 sequential acquisition 36 SHIFMN 137 SHIFMX 137 shift priors 148 shift referencing 136 correction 148 149 sign convention 98 Siemens data 26 41 59 125 signal to noise ratio
95. ON 9 3 1 2 One spectrum Single spectrum with no water suppression The Two Spectra Method is usually preferable In principle this One Spectrum Method is attractive because the additional water peak allows better phasing ref erencing and consistent resonance area ratios to water However it should only be used if the lipid peak around 1 3 ppm is usually at least 25 the height of the water peak Otherwise the water signal can dominate the analysis at the expense of the accuracy of the other metabolites especially the weaker ones You acquire a single spectrum without water suppression with as many scans as you usually use for water suppressed spectra Alternatively with GE Probe data you can use Sec 7 4 5 1 to analyze only the many unsuppressed data frames using the Preprocessor use unsuppressed documented in HOME 1cmodel gelx preprocessors use unsuppressed For consistency in a study it is important to always use exclusively the One Spectrum Method of this section or the Two Spectra Method of Sec 9 3 1 1 not both With the One Spectrum Method you also must input PPMST 8 0 If your spectrum does not go up to 8 0 ppm i e you get the diagnostic FATAL INITIA 3 decrease PPMST to be about 0 3 ppm inside the spectral range However keep PPMST gt 7 0 to fully include the water peak and enough adjacent baseline 9 3 1 3 Excising residual water If you also want to estimate Lip53 Lip52 Sec 9 3 2 you should u
96. Otherwise do not do this The resolution suffers and much worse this nonlinear transfor mation introduces errors the absolute value of a sum is not the sum of the absolute values Information is destroyed and artifacts introduced by taking absolute values of overlapping signals For Bruker users only There is no need to use magnitude spectra You correct the shift caused by Bruker s digital filter with Bruker s convdta or automatically with LCMgui Sec 7 3 1 The 1st order phase correction is then moderate ABSVAL LOGICAL TRUE to analyze the absolute value of the data You input the usual complex RAW and BASIS files and LCModel does the conversions Do not input absolute values Default ABSVAL F 11 17 Consistency with Old Versions This is only for users who want consistency with Versions 5 x e g of concentration estimates in longitudinal studies or of output format for your own post processing programs SPTYPE version5 sets the following NRATIO O NSIMUL 0 PPMEND 1 0 PPMST 3 85 The results will then be closer but not identical to Version 5 x results Contrary to the other SPTYPE settings it is not possible to overwrite these Control Parameters If you want to change only some of the four above Control Parameters then input them and do not input SPTYPE Other settings for backward consistency are NTITLE INTEGER Input NTITLE 1 for the old one line TITLE of Version 5 x Default
97. RAW file is needed for each metabolite in the Basis Set All of the Control Parameters for these two types of files are specified in Secs 8 6 3 amp 8 6 4 However we will first discuss the general features and usage of MakeBasis with examples often mentioning Control Parameters that are not defined in detail until later in Secs 8 6 3 amp 8 6 4 MakeBasis does the following e optionally phases your Basis Spectra Auto Phasing e references your Basis Spectra e optionally consistently scales your Basis Spectra with each other Auto Scaling e outputs the BASIS file that will be used by LCModel To do this MakeBasis requires the formate standard for Auto Phasing and Auto Scaling and the DSS marker for Auto Phasing and referencing as specified in Sec 8 2 4 Each spectrum is scaled multiplied by the Control Parameter TRAMP in its RAW file The metabolite spectra used for the test run in Chap 4 were acquired before Auto Scaling was available and have no formate standard Auto Scaling and Auto Phasing 8 6 RUNNING MAKEBASIS 91 therefore had to be suppressed by setting AUTOSC AUTOPH F in makebasis in Thus the phase corrections and the TRAMP scale factors had to be carefully determined beforehand and input 8 6 1 Auto Scaling Auto Scaling and Auto Phasing are illustrated with a small set of test data provided with your package Enter cd lcmodel bin makebasis lt 1cmodel test control makebasis auto in Th
98. SA LABpub 2 246 eet Baws ine Ole Meth bet bs wae lie Se 107 973 2 OUTPUT aoga ha a a ai we e Bea ok eee ee 109 9 3 3 Estimating only choline 110 9 4 MEGA PRESS for GABA 0 2 00000 0005 111 YAA MPU nie i tice A E Gee eee ta Mee oi 111 DAD I OUTPUT Jes tion keh wos a da OF ees Oo E a 114 9 5 Internal Logical Units o 0 0 00 020004 114 9 6 Eddy Current Correction 2 0 0 a 115 9 7 Prior Phasing Information e 115 9 7 1 Estimating Appropriate SDDEGZ SDDEGP DEGZER amp DEGPPM 116 9 8 Specifying the Basis Spectra for the Analysis 117 9 8 1 Omitting Basis Spectra from the Analysis 117 9 8 2 Keeping Basis Spectra in the Analysis 117 9 9 One Page Output 118 9 9 1 Concentration Table 0 o e 118 9 9 2 Diagnostics Table 2 2 0 0 0 2 eee ee 120 US BLOG BB et 120 9 9 4 Miscellaneous Output Table 122 10 Absolute Metabolite Concentrations 124 10 1 Calibration Phantoms e 125 10 1 1 Data Scaling intra hardware lt lt 125 10 1 2 Scanner Calibration inter hardware 127 102 Water SCaline snit ai onti es BE OES Se Ps 128 10 28 Method tirerai kas te eee Date wer beet 128 CONTENTS 10 22 US LE cass di Rein ee Bye aah agree Bak es eens 11 Fine Points 11 1 Error Estimates amp Reproducibility 11 1 1 Ave
99. Siemens Vision or 1 5T GE unless you can do your own scaling for instrumental effects as in Sec 10 1 1 e For data where voxel size is uncertain e For convenience and simplicity 10 1 Calibration Phantoms The Scale Factor for scaling data consistently with each other actually has two factors intra hardware to account for differences in acquisition conditions e g coil load ing with the same hardware that change the total effective gain Sec 10 1 1 inter hardware to scale the data with the Basis Set to account for hardware changes or different hardware Sec 10 1 2 10 1 1 Data Scaling intra hardware Each data set RAW file Sec 5 2 2 is scaled by the factor TRAMP VOLUME LCMgui computes VOLUME as the VOI volume in mL from the GE Marconi Picker Philips amp Siemens headers TRAMP must then be inversely proportional to the total gain accounting for varying acquisition conditions so that a given number of protons always produces the same resonance area in each spectrum 10 1 1 1 Siemens Data The following only works with Siemens Vision data TRAMP is set equal to the RF transmitter amplitude of a non selective 90 reference pulse rectangular and 0 5 ms 126 CHAPTER 10 ABSOLUTE METABOLITE CONCENTRATIONS in duration 17 For Siemens Vision data LCMgui sets TRAMP to the raw file header element G19 Acq3 Mr TransmitterCalibration You cannot use this with other Siemens data The above TRAMP is me
100. T 16 is the analysis of the same voxel with PPMEND 1 8 In this particular case the concentration ratios do not change drastically but they can be unreliable in cases like PLOT 15 In these cases you should input PPMEND 1 8 For brain never set PPMEND lt 0 1 The model metabolite spectra usually have a DSS marker around 0 0 ppm that must be kept out of the analysis For brain PPMEND should never exceed 1 9 3 4 2 PPMST PPMST 4 0 seldom has to be changed Only when the noise is much higher in the region near 4 0 ppm would a lower PPMST be called for If this would require a PPMST lt 3 85 then reject the spectrum or try omitting eddy current correction LCModel attempts to find the smoothest Baseline that is still consistent with the data No other restrictive assumptions are made In this sense it is an efficient filter in frequency space It can handle the tail of the residual water signal better than digital post processing with more restrictive assumptions and consequent distortions Do not use too strong water suppression which can attenuate metabolite signals below 4 ppm Even with moderate water suppression it is usually best to leave PPMST 4 0 at least at fields below 4 7T 22 CHAPTER 3 ESSENTIAL GUIDE 3 5 Criteria for Rejecting Analyses 3 5 1 Non Random Residuals The residuals are the data values minus the fit values They are plotted at the top of the plot in the One Page Output They should appear rand
101. T session you can exit normally by simply clicking Exit amp Save in Fig 6 6 7 2 5 Install License 7 2 5 1 Linux You receive your license file attached to an email You install it simply with the command mv license HOME 1lcmodel If your email comes through some other system it might rename the file license dat or license bin Be sure that the file is correctly renamed back to license be sure to transfer license to your Linux computer using binary mode not ascii and not text mode 7 2 5 2 Sun SGI or DEC Compaq After you have completed the test runs a window appears with an Install License button Clicking this produces the window in Fig 7 1 You enter the information that you received with your license After clicking OK and accepting the License Agreement from the website 7 2 INSTALLING LCMGUI 57 on page 2 the File Selector Sec 6 2 1 appears and you are ready to start the first analysis of your data see Sec 7 3 be sure to exit normally from LCMgui with the Exit amp Save button in Fig 6 6 so that your license information is saved 7 2 5 3 GE SAGE LCModel interface With Linux set the following in your sagerc file setenv LCM_OWNER dummy setenv LCM_RUNKEY 123 They are not used by the Linux version With Sun SGI or DEC Compagq replace dummy above with the OWNER string provided with your license and replace 123 above with the value for KEY provided
102. TAL When using only simulated basis spectra as with most values of SPTYPE you cannot have BASCAL T NCALIB gt 0 or NSIMUL lt 0 FATAL Error trying to open FILBAS Hints Check for non existent file or permissions FATAL Error trying to read Namelist BASIS1 in FILBAS Hints See Sec 12 1 1 FATAL Error trying to read Namelist BASIS in FILBAS Hints See Sec 12 1 1 FATAL Error trying to read a Basis Spectrum in FILBAS Hints In FILBAS check that the Basis Spectrum data correspond to FMTBAS Not used FATAL You have not input anything for FILBAS or you input LBASIS lt 0 ERROR It was not possible to scale the simulated peaks They will all differ from the concentrations of the metabolites in the BASIS file by an unknown factor Hints By default the Cr from the BASIS is used for scaling Does your BASIS file have creatine with the Metabolite Name Cr For example if it is instead Cre then you must input WSMET Cre If you use something other than the Cr CH singlet for scaling then you will also have to input WSPPM amp N1HMET FATAL There is not enough storage to convert the basis set to the bandwidth of your data Hints Contact me about a modified basis set or data or about a larger version of LCModel FATAL With SPTYPE lipid 8 breast 8 or liver 11 you must have the Hidden Control Parameter NOBASI T and CHSIMU must have a special undocumented structure Hints Do not use CHSIMU
103. TANDARD LCMODEL DIAGNOSTICS 155 ERROR an error condition has occurred but corrective action has been taken the results could be affected FATAL an error that cannot be corrected has occurred the run is aborted ILLOGICAL these diagnostics were used during development and testing to detect programming errors they should never occur There are also the following special info messages Doing Water Scaling Water Scaling has been done CRUDE MODEL The results are based on the Preliminary Analysis with the crude Gaussian Lorentzian model and the analysis with the full model has been suppressed by your Control Parameters In the following all LCModel diagnostics are listed alphabetically and explained The diagnostic name also identifies the subprogram and the general purpose of this subprogram and general hints are sometimes given Read this before going to the diagnostic number Then the individual diagnostics are listed by number Possible remedies and suggestions are given after the word Hints There are no Hints with the ILLOGICAL type or when the remedy is obvious There are standard Hints with the following meanings Never Happened This diagnostic is unlikely to occur Therefore further hints are usually not given Working Hard LCModel performs many independent analyses typically about 100 from different starting points and usually recovers completely from difficulties in one of these analyses This diagnostic h
104. Varian files in Sec 3 7 which are automatically handled by LCMgui 5 3 CONTROL File An example from the test runs is the file test control The CONTROL file consists solely of the the Namelist LCMODL A simple example is LCMODL OWNER Biomedizinische NMR Forschungs GmbH KEY 123456789 TITLE pat_777_77 41Y M 18 0m1 gm PMP FILBAS home 1cm basis sets te20_2c basis FILRAW pat_777_77 raw FILPS pat_777_77 ps END The above Control Parameters are OWNER CHARACTER 120 the name of the license holder With Linux this is not used but the Linux license file must be in HOME 1cmodel license This string is given to you with your license You must input it exactly i e with the same capitalization and spacing all on one line KEY INTEGER 20 your license key With Linux this is not used but the Linux license file must be in HOME 1cmodel license This is given to you with your license Together with OWNER it enables you to run LCModel on the licensed Unix workstation TITLE CHARACTER 244 the title that you want to appear at the top of the One Page Output and elsewhere FILBAS CHARACTER 255 the pathname of the BASIS file to be input Note On most platforms you cannot use symbols like or HOME to denote your home directory with any Control Parameters FILRAW CHARACTER 255 the pathname of the RAW file to be input FILPS CHARACTER 255 the pathname of the PS file th
105. View Edit Control Parameters and enter the above Control Parameters including HZPPPM etc listed above in the window in Fig 7 7 Click OK and then Run LCModel e If you wish you can save these changes e g in bascal gpc for possible future use e When you exit LCMgui you can click on Exit amp Forget This will cause your Other file type to be forgotten and your next LCMgui session will start as your present one did expecting the usual file type Finally the definitions of the above Control Parameters are BASCAL LOGICAL Basis calibration TRUE to analyze one spectrum of the Basis Set with another spectrum of the Basis Set Note only spectra in the Basis Set can be used not spectra simulated with NSIMUL which is set to zero Default BASCAL F CHBCAL CHARACTER 6 is the Metabolite Name of the Basis Spectrum that will 100 CHAPTER 8 MAKING THE BASIS SET be analyzed as data when BASCAL T NCALIB INTEGER NCALIB gt 0 specifies that only NCALIB Basis Spectra from the Basis Set will be used for the analysis The normal use is NCALIB 1 Default NCALIB 0 which will cause FATAL MYBASI 12 when BASCAL T i e you must input NCALIB when BASCAL T CHCALI CHARACTER 60 6 The first NCALIB elements contain the Metabolite Names of the Basis Spectra that will be used in the analysis In addition these Basis Spectra will automatically be used in the Prelim inary Analy
106. W and gaba h20 are LCModel RAW files with the time domain data for the metabolite with the normal protocol and with the Unsuppressed Data re spectively These data files for KECC have exactly the same meanings and formats as those specified in Sec 9 6 Section 5 3 1 specifies NUNFIL In case of an abort a self explanatory error message is written at the end of the ecc file 8 5 PLOTTING RAW FILES WITH PLOTRAW 87 8 5 Plotting RAW Files with PlotRaw The program PlotRaw can plot the time domain or frequency domain data directly from a RAW file It is useful for choosing input parameters for MakeBasis as well as for looking at suspect in vivo data in detail You could also skip Sec 8 5 and use the plots of the spectra output by MakeBasis The only advantage of PlotRaw is that you can also plot the time domain data and absolute value spectra You should first plot the absolute value of your model metabolite spectra with PlotRaw All plots are properly scaled by your factor TRAMP VOLUME PlotRaw is also implemented in LCMgui You simply click on Preview Data in the main Control Parameters window in Fig 7 2 If you want to plot a RAW file select the RAW file click Retry in Fig 6 2 click Other in Fig 6 3 and then select your RAW file again LCMgui users can skip the rest of Sec 8 5 8 5 1 Test Run of PlotRaw For this test the file HOME 1cmodel test control plotraw in has been supplied to you You enter
107. Z are automatically set to exploit this Saving these as defaults and using them with data of other vendors can result in disastrous phasing errors With other data check Fig 7 7 and delete them if necessary 3 7 VENDOR SPECIFIC GUIDELINES 25 3 7 3 Hitachi e LCModel is part of the Hitachi system There is nothing that you need to do 3 7 4 Marconi Picker e LCMgui handles single voxel spectroscopy fid dump files obtained by clicking the dump button in your Marconi Picker software e If you are using PRESS then see Sec 7 3 1 using LCMgui or Sec 5 3 5 not using LCMgui 3 7 5 Bruker Philips Siemens Toshiba Varian Microsoft Do not use spaces in filenames or directory names Transfer your data files directly to a Linux or Unix system With FTP transfer with binary mode not text and not ASCII mode Avoid file manipulations on Microsoft amp Apple systems as much as possible They can introduce non printing characters that make the files unreadable on other systems MATLAB amp IDL have reportedly done this as well 3 7 6 Philips e LCMgui handles single voxel and CSI DICOM and SDAT files containing the time domain data LCMgui does not accept XSDAT files e DICOM files If you get an abort with the error message that the file is not a complete spectroscopy file you must set the Philips output export configuration to Enhanced MR not Classic MR You only have to select t
108. acquired for each in vivo spectrum and subtracted from it 33 This reduces the S N but as in 1 any errors in subtraction e g due to changes during acquisition of the two spectra would often require smaller corrections by the Baseline than without subtraction 3 Additional simulated lipid and or MM spectra can be added as Basis Spectra 5 3 Approach 3 is used here It requires a difficult delicate balance between over parameter 11 8 CONCENTRATION RATIO PRIORS 143 ization and over simplification Regularization provides this balance for the Baseline amp line shape but the concentrations themselves are normally not regularized and there are now far too many concentrations to estimate without some type of stabilization The Concentration Ratio Priors below attempt to do this LCModel uses the following Parsimony In 5 3 the simulated components were parsimoniously chosen so as not to over parameterize the analysis Similarly MM20 combines three components around 2 25 amp 2 0 ppm and also includes a weak component near 3 0 ppm Allow ing these four components to be independent produced flat Baselines but great scatter and unrealistic values in concentrations in test sets of control spectra Similarly Lip20 is a combination Lineshape Concentration estimates depend critically on the lineshape model Normally all the resonances in the Analysis Window are simultaneously used to determine the nearly model free regu
109. ade runs from your account then you may not have to change any settings and the following three simple steps may be sufficient e Start LCMgui Sec 6 1 e Select the file containing the data to be analyzed Fig 6 1 in Sec 6 2 1 e Click on Run LCModel Fig 6 4 in Sec 6 4 Parts in this fine print are seldom necessary reading but they can help you make better use of LCMgui If you are one of the first to run LCMgui from this account then you should check the points in fine print 6 1 Starting LCMgui LCMgui is installed under one or more usernames You must log on using one of these usernames Log in normally as the user not as root You can then always start LCMgui with HOME 1cmode1 1cmgui LCMgui may have been installed so that you can also start it with simply 1cmgui 6 2 Selecting the Data to be Analyzed See Sec 3 7 for a list of the types of data files automatically handled by LCMgui and for important guidelines 44 6 2 SELECTING THE DATA TO BE ANALYZED 45 a Select your GE Signa LX Raw P file G X Directories J Files o home sp tmp chla 0210 S p7 P 7 P 7 gt 3 Cancel 4 Show Hidden Files p urong type 7 List Files of Type Fr Figure 6 1 The File Selector shown here for GE LX files 6 2 1 File Selector Window When LCMgui is started the first window that appears is usually the File Selector An example with GE Signa LX data is shown in Fig 6 1 With d
110. aling of the simulated metabolites will be inconsistent with the unknown scaling of the metabolites in the Basis Set 2 2 PLOT 17 2 1 3 Prior Ratio Probabilities Section 11 8 shows how prior probabilities so called soft constraints are imposed on ratios of some poorly determined concentrations to more accurate concentrations An example is Lip09 Lip13a Lip13b the ratio of the lipids near 0 9 ppm to those near 1 3 ppm In PLOT 1 in HOME 1cmodel doc figures pdf the line with Lip09 is bold italic red and the name Lip09 is preceded by a minus sign This means that the Lip09 concentration is significantly lower than expected A plus would mean significantly higher This is common with the highly variable lipids and macromolecules There are similar very weak constraints on the weak normal metabolites Asp GABA Glc Scyllo and Tau If one of these is output in red italic then consider whether this may be a real effect due to a pathology If so then use Sec 11 8 1 to omit this constraint 2 2 Plot The real part of the frequency domain data spectrum i e the phased and referenced FFT of your raw input data with no smoothing is plotted as a thin curve The thick red curve is the LCModel fit to this data Also plotted as a thin curve is the so called Baseline We will call the ppm range of the plot the Analysis Window i e the range over which the data are analyzed At the top are plotted the Residuals i e the data mi
111. an specify prior probabilities soft constraints on concentration ratios as follows NRATIO INTEGER is the number of soft constraints on concentration ratios that are spec ified Default NRATIO 12 i e CHRATO 13 below is not used CHRATO CHARACTER 60 264 specifies the soft constraints Defaults CHRATO 1 Lip09 Lip13 0 267 0 1 WT MM12 CHRATO 2 Lip20 Lip13 0 15 0 07 WT MM12 CHRATO 3 MM20 MM09 1 5 0 375 WT Lip09 CHRATO 4 MM12 MMO9 0 3 0 1 WT Lip09 CHRATO 5 MM14 MMO9 0 75 0 45 WT Lip09 CHRATO 6 MM17 MMO9 0 375 0 3 WT Lip09 CHRATO 7 CrCH2 totCr 0 1 0 15 1ooor il 144 CHAPTER 11 FINE POINTS CHRATO 8 Asp Big3 0 05 0 05 CHRATO 9 GABA Big3 0 04 0 04 CHRATO 10 Glc Big3 0 05 0 075 CHRATO 11 Scyllo Big3 0 02 0 02 CHRATO 12 Tau Big3 0 04 0 06 CHRATO 13 NAAG NAA 0 15 0 15 These are of the form metnum metdenom expect stddev WT metwt metnum the numerator of the ratio it must be a single Metabolite Name not a Concentration Sum metdenom the denominator The more accurate concentration in the ratio should be in the denominator It can also be a Concentration Sum There are also the following special symbols A wildcard at the end will match any string of zero or more charac ters With the default Metabolite Names in Sec 11
112. and GPC are expensive but do not go below their minimum concentrations in Table 8 1 8 2 1 1 Singlets Glycine No concentration values are given for metabolites that only have one singlet in the Analysis Window You simulate these using Sec 11 7 where simulating Scyllo is illustrated Inputting NSIMUL 14 to LCModel will simulate Glyc Sec 11 7 Normally you do not use Glyc it is poorly resolved from Ins at 3 56 ppm Worse Ins has a different effective number of protons 17 which would cause unnecessary scatter even in the sum Ins Glyc about 15 if Glyc is mistaken as Ins or vice versa As an exception 80 CHAPTER 8 MAKING THE BASIS SET when Glyc is pathologically high e g in some tumors you might want to replace Ins with a simulated Glyc Sec 9 8 1 especially at long TE where Ins is weak In these pathologies even at short T E you might as an exception use both Ins and a simulated Glyc If you do use both Ins amp Glyc you could input PPMST 4 2 if the baseline has no sharp increase or decrease above 4 0 ppm The Ins signal around 4 06 ppm should then help to resolve the two 8 2 1 2 NAAG NAAG is expensive and one compromise is to use the NAA spectrum shifted with the main peak shifted to 2 045 ppm from 2 01 ppm This is a good approximation to NAAG except that a small NAAG peak near 2 15 ppm is missing Compare PLOTs 3 amp 4 in HOME 1cmode1 doc figures pdf This can lead to reduced NAAG NAA estimates i
113. aningless in the usual case that the transmit and receive coils are not the same For non Vision files TRAMP is simply set to 1 0 Also VOLUME is not correctly obtained for multi voxel DICOM files use rda files And of course headerless raw files have no such information So you would have to use Water Scaling or a more elaborate calibration procedure 23 10 1 1 2 GE data The following formulas are used for all GE data but they reportedly do not accurately describe the behavior with TG at 3T So Calibration Phantoms should only be used at 1 5T With 3T data and all phased array data you would have to use Water Scaling Sec 10 2 With GE Signa 5 x systems LCMgui uses TRAMP 2 6 R1 30 R2 11 0 005 TG 65 where I have arbitrarily chosen R1 6 R2 30 and TG 65 to be the reference point where TRAMP 1 0 This reference point is unimportant since it is consistently used for all TRAMP With GE Signa LX data R1 is replaced by R1 2 TRAMP 2 6 R1 2 30 R2 100 005 TG 65 10 1 1 3 Data Other than Siemens Vision or 1 5 T GE Except for Siemens and GE data LCMgui simply sets TRAMP 1 0 Therefore you would have to do one of the following e Use Water Scaling Sec 10 2 e Compute your own TRAMP e g using a more elaborate calibration procedure 23 10 1 1 4 Fine Points In addition to 17 there are detailed literature references studies alternatives and refinements to determining TRAMP
114. archiving these files You can usually print the file from your display program anyway 4 2 Test Runs Using LCMgui 4 2 1 Display on Your Screen Try just clicking the Continue button in Fig 4 1 to use the default evince which is usually installed If it is not LCMgui will give you a chance to enter another display command e g gv or okular in the green field You can later download and install 28 4 2 TEST RUNS USING LCMGUI 29 a Select PostScript Display or Print Command UE Welcome to LCMgui the graphical user interface to LCModel LCModel LCMeui have not yet been fully installed on this account The installation amp Test Runs are starting The LCModel test results are summarized on 4 pages usually only 1 or 2 pages of PostScript output which are automatically displayed by the command in the green field below First try clicking the Continue button below to use the default display command evince If this fails see Chapter 4 of the manual HOME lomodel doc manual pdf gt for installing evince or entering an alternative display command e g okular in the green field below You can also exit LCMgui and use the Linux command ps2pdf or Adobe Acrobat Distiller to produce a PDF file The above displays allow printing so a print command typically lpr gt below is usually not necessary Enter display or print command vine Continue Abort LCMgui Figure 4 1 Here you enter your com
115. ard PNNLS partial nonnegative least squares does a quadratic programming analysis at each iteration in the nonlinear least squares analysis 1 info Maximum iterations before convergence in the quadratic program corrective action is taken Hints Working Hard This is common when there is only one dominant peak as often with lipid dominated spectra such as muscle or liver It is especially critical when prior phasing information has not been input using SDDEGZ or SDDEGP It can occur many times in one analysis With prostate or lipid dominated spectra it does not seem to be a problem PPRIOR parses the CHRATO for the Concentration Ratio Priors The error number is JPRIOR 100 x TERR which are listed below For example if you get FATAL PPRIOR 213 then your CHRATO 13 caused the diagnostic labeled 2nn below All PPRIOR errors are FATAL Hints See Sec 11 8 for the symbols below The tokens and rules in Sec 11 8 must be followed exactly The errors here refer to wrong tokens or format The actual values are checked elsewhere The number of spaces between elements should not be important but only spaces not tabs etc are allowed nn CHRATO nn is an empty string or the token is missing or 168 RBASOL RFALSI CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS metnum metdenom is missing lnn expect is missing or non numeric or the token is missing 2nn stddev is missing or non numeric or the token WT is mis
116. as Lac and distort the referencing The elements of PPMREF JREF 2 should increase with JREF since NREFPK 2 is automatically decremented if problems at high ppm due to incomplete water suppression are detected SHIFMN REAL 2 SHIFMN JCCF is the minimum in ppm allowed endpoint for the referencing shift grid obtained from CCF JCCF Default SHIFMN 0 2 0 1 SHIFMX REAL 2 SHIFMX JCCF is the maximum in ppm allowed endpoint for the referencing shift grid obtained from CCF JCCF Default SHIFMX 0 3 0 3 NSHIFT INTEGER the maximum number of different starting points from the referencing shift grid The referencing shift grid itself is much finer Each new starting point is the grid point farthest from all preceding parameter trajectories You must have NSHIFT lt 25 Default NSHIFT 8 11 4 Multi Voxel Data Sets 11 41 GE CSI data For GE CSI data sets use the Research Version of SAGE which now has Mary McLean s LCModel interface 9 When using this SAGE LCModel interface a window titled LCModel Analysis appears See Sec 7 2 5 3 for important necessary settings and discussion Section 3 4 1 discusses cases where you might want to set other values for End PPM PPMEND 11 42 Non GE CSI data Most other 2D amp 3D CSI data sets can be handled by LCMgui See Sec 3 7 for details For other types of CSI data you can make your own LCModel RAW file as specified in Sec 5 2 3 1 With LCMgui or PlotRa
117. as occurred in one of these many analyses and will probably not affect the final results How ever these diagnostics are useful signals that that LCModel is working hard on difficult data You can judge from the One Page Output if your results have been affected Check that the fit to the data is reasonable and the other criteria discussed in Sec 3 5 are satisfied Control Parameters and other terms are usually not cross referenced The Index at the end of this manual tells you where you can locate their definitions AREABA computes the resonance area of the reference Basis Set singlet specified by WSMET for Water Scaling Errors here simply cause the analysis to proceed without any scaling Only the concentration ratios of the metabolites in the BASIS file will be valid If you are simulating Basis Spectra inside LCModel Sec 11 7 then the concentrations of all these simulated metabolites would have to be multiplied by an unknown scale factor to be consistent with the concentrations of the BASIS metabolites All AREABA diagnostics are of type ERROR 1 The integration range for the Basis Set singlet Cr at 3 03 ppm by default used for scaling is not completely in the Analysis Window Hints Extend your Analysis Window with PPMST or PPMEND or choose 156 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS 2 3 another singlet with WSMET etc One of the following was not input positive N1HMET ATTMET The integral of the refere
118. ase the resolution of your display Otherwise you can slightly reduce the windows by reducing 10 in i_fontPoints 10 Missing fonts in your Linux Unix distribution may cause the characters to be poorly displayed Changing i_fontPoints 10 above may help although the best collection of fonts is often 10 point 7 5 1 5 Help Balloon Delay Particularly in the Control Parameters window balloons with help messages pop up after the cursor has been over a field for i_balloonWait 3000 ms A smaller value may be useful to beginners but irritating to experienced users 7 5 FINE POINTS TT 7 5 1 6 Colors The five colors specified in gui defaults all have self explanatory indices beginning with i color For example i_colorFollowMe yellow specifies that the normal path to take is denoted by yellow buttons You can change these values to symbolic names e g red or by hexadecimal numbers preceded by a e g ff0000 with the RGB value Your X Windows system must be able to assign the color This manual will then no longer be correct in its color references to buttons and fields 7 5 2 Cleaning up the File System 7 5 2 1 Obsolete User Defined Files You can produce and save arbitrarily many files with Control Defaults Preprocessors and Execution Scripts They are saved in HOME 1cmodel profiles control defaults HOME 1cmodel execution scripts HOME 1cmode1 77 7 preprocessors where is your User Profile and bruke
119. at will contain the PostScript One Page Output You will have to add more changes to your CONTROL file These are specified in this chapter However these changes are usually always the same the only Control Parameters in the CONTROL file that are normally changed with every data set are 38 CHAPTER 5 RUNNING LCMODEL WITHOUT LCMGUI BASIC INPUT TITLE FILRAW and FILPS You should be able to write a simple Pre Formatting Program so that the Normal User only has to input one name e g pat_777_77 The Pre Formatting Program then e finds the corresponding raw time domain data and formats it as the RAW file e makes the CONTROL file by extracting TITLE from a header in your original data file and by inserting pat_777_77 into FILRAW and FILPS e runs LCModel and prints or displays the One Page Output in the PS file In the following I specify the additional generally data set independent changes that your Pre Formatting Program adds to the CONTROL file to change the default Control Parameters to those appropriate for your instrument and protocol Note that the ordering in a Namelist is not important except that a Control Parameter input several times will take on the value closest to the end of the Namelist It is therefore best to add changes to the bottom of the list just above END The table Input Changes in the One Page Output prints your CONTROL file from the bottom up so that the most recent changes at the bottom of the
120. ata Clicking Preview Data at the bottom of the Control Parameters window Fig 7 2 produces the window in Fig 7 13 As explained there by editing the green field next to Print the plot with command to your command for displaying or printing PostScript files and then clicking on Run Preview you can display or print plots of the data before starting LCModel This can be useful in checking whether you have selected the correct data file and whether the data are worth analyzing y You can activate both the Display or print amp Second display buttons to both print and display 70 CHAPTER 7 LCMGUI REFERENCE MANUAL EJ Preview Data oo EE You can change any parameters below When you are done click on Run Preview below Note that in contrast to LCModel this Preview does not reference the data However it will attempt an eddy current correction with GE P files and Marconi Picker files Plotting data from 4 0 ppm down tos 0 2 ppm M Plot absolute value of spectrum W Plot time domain data M Plot spectrum with zero order and first order phase corrections Po degrees and bo deg ppm M Display or print the plot with the comand A Second display or plot command IBIS ttt Run Preview Cancel Preview and Continue Figure 7 13 Here you can set and execute your commands for printing and or plot ting previews of your data If possible orientation and magnificatio
121. ata from other manufacturers the title Select your GE Signa LX will be different LCMgui remembers the directory where the data file was in the last run and the File Selector starts there If you have several User Profiles then the window in Fig 7 12 will appear first Normally you immediately click on Use Selected Profile to continue using your present User Profile If the last LCMgui session did not use Bruker GE Marconi Picker Philips Siemens Toshiba or Varian data then the window in Fig 6 3 will appear first You normally then click the Other button in Fig 6 3 as you did in the previous session If no license has been installed a window will first appear offering you the chance to test the conversion of your data as in Sec 4 2 4 You first navigate through the directory tree When you click on a directory its files are listed in the right pane the stage shown in Fig 6 1 You then click on the filename that you want to use and click the yellow OK button at the right or simply double click on the filename Only filenames that match the filter shown at the top of the pane of filenames p 7 P 7 or P 7 in Fig 6 1 are displayed If your filename differs from this standard e g if you analyzed GE Signa 5 x data last time and LX data this time then it will not be listed You can change 46 CHAPTER 6 ELEMENTARY GUIDE TO LCMGUI ed Wrong file type Try again The following f
122. atal error s occurred while trying to convert your GeLx file tmpichla 02 10 p wrong type 7 Error reading start of header in tmp chla 0210 p wrong type 7 Is this file of the correct type You probably selected the wrong type of file Click on Retry below to try selecting again Retry Abort LCMgui Figure 6 2 You selected a file of the wrong type in the File Selector the filter by clicking on the field p 7 P 7 Px 7 pressing the backspace or delete key entering your new filter e g and pressing the Enter key If you want to also see files or directories that start with a dot e g 1cmode1 then click on the Show Hidden Files button on the right to make it yellow If the filenames or directories in Fig 6 1 are so long that they are cut off then the width of the panes can be adjusted by holding the left mouse button down on the small square at the top on the pane boundary between Directories and Files and dragging the pane boundary You can also simply use the scroll bar that appears at the bottom of the pane when names are too long 6 2 1 1 Selecting the wrong file type LCMgui remembers what type of data was analyzed from this account and User Profile the last time and it expects the same type this time You therefore get a window like Fig 6 2 if in the File Selector you e Select a file of another type e g a Siemens raw file after GE
123. bolite Names of the spectra to be kept in the analysis For example NKEEP 1 CHKEEP 1 Lac in the CONTROL file would keep Lac in the current analysis even if PPMEND gt 1 41 not recommended 9 9 One Page Output 9 9 1 Concentration Table 9 9 1 1 Concentration Sums It is important to have separate Basis Spectra for all metabolites that have a chance of being resolved even if they are poorly resolved e g NAA and NAAG However the sum of the concentrations of metabolites with similar spectra can be much more accurately determined than the individual concentrations and it is important to also output these sums together with their SD values in the concentration table NCOMBI INTEGER no of combinations the number of sums that you want output in the Concentration Table Default NCOMBI 16 CHCOMB CHARACTER 100 56 the first NCOMBI elements specify the metabolite concentrations to be summed as a string with the Metabolite Names sep arated by with no spaces There can be up to six Metabolite Names in the sum Although long CHCOMB may be truncated in the Concentration Table the full sums will be used Default CHCOMB GPC PCh GPC Cho PCh Cho Cho GPC PCh NAA NAAG Ins Glyc mI Glyc Ins Gly mI Gly Cr PCr gt Cre PCr GlutGln Lipi3a Lip13b MM14 Lip13a Lip13b MM12 MMO9 Lip09 MM20 Lip20 For example Thr amp
124. cant uncertainties in the corrections needed for this Chapter 10 also discusses an alternative for Siemens Vision and 1 5T GE data using Calibration Phantoms You activate Water Scaling simply by making the Do water scaling button blue However Sec 10 2 2 specifies important additional Control Parameters that you may have to change from their default values Sections 7 3 4 amp 7 3 5 specify how to change Control Parameters and save these changes for future use 6 3 CONTROL PARAMETERS WINDOW 49 Concentration ratios are not affected by Water Scaling and are less sensitive to relaxation and partial volume effects 6 3 5 Only for Multi Voxel data files If you only have single voxel data then you can skip this section and this part of the Control Parameters window The balloon that pops up over this part gives step by step instructions on previewing a whole slice and then selecting a rectangular subset to be analyzed It also gives troubleshooting hints Some headers do not have complete information on the total numbers of slices rows amp columns in the data file even though LCMgui still correctly converts the file So you should first check and if necessary correct these values usually signaled by a i The Preview Data button at the bottom of the Control Parameters window is especially important with CSI data sets Details and troubleshooting hints are given in Sec 7 4 1 6 3 5 1 Multi Voxel out
125. cessively long acquisition times per scan n 1 and Ns Ng 9 9 ONE PAGE OUTPUT 123 6 infls the number of inflection points in a typical lineshape extrs the number of extrema in this lineshape Numbers in parentheses are the corresponding values at an initial stage The concentrations are determined by an integral over the lineshape and this is a well determined quantity even if the lineshape itself has several extrema 19 Chapter 10 Absolute Metabolite Concentrations You only need to read this chapter if you want to estimate absolute concentrations Concentration ratios are not affected by any of the settings in this chapter Note that in addition to the scaling correction discussed in this chapter you must also correct for relaxation Sec 11 2 and partial volume effects Sec 10 2 2 3 Con centration ratios are less sensitive to these effects you should consider ratios at least as supplemental data The LCModel concentrations estimate how much of each Basis Spectrum is in the in vivo data Since the Basis Set and the in vivo data are generally not acquired under identical conditions scanner coil loading etc their scaling is different and the LCModel concentrations will all be off e g from mM by an unknown Scale Factor Therefore for meaningful absolute concentrations the in vivo data must be first scaled by this factor so that they are consistent with the Basis Spectra To determine this Scale Factor
126. coni formerly Picker now Philips Versions 9 x 10 2B dump files Philips e SDAT files Siemens rda file transferred from syngo PC or Vision raw file from old Sun console Toshiba DICOM 47 xx 2 or rawData V4 10 6 30 files Varian fid Files Other including versions other than above such as the old Siemens SP REEEREEE Figure 6 3 Note that GE Signa 5 x and LX or higher have distinct buttons However you should check the following in your Control Parameters window at least during your first runs Most of the fields have balloon help messages that pop up if the cursor remains over a field for 3 seconds 6 3 1 Analyzing spectrum from In the second row in Fig 6 4 the entries next to Analyzing spectrum from define the Analysis Window the ppm range over which the spectrum will be analyzed Sec 3 4 You can change this range by clicking on the green up or down arrow next to the value Green usually indicates fields that you can click on or edit to make changes You may want to change these values to cut off artifacts Sec 3 6 1 6 3 2 BASIS file In the third row in Fig 6 4 the entry next to BASIS file shows the pathname of the BASIS file containing the model spectra that will be used to analyze your data If the BASIS files are installed as specified in Sec 7 2 4 then LCMgui should automatically select the proper one if it exists Also LCModel normally produces diagnostics if
127. defaults when you input values yourself The Control Parameter SPTYPE Sec 9 1 informs LCModel of special types of spectra remember that PLOTs are in HOME 1cmodel doc figures pdf e Weak NAA Tumors Lesions Input SPTYPE tumor when you suspect that the NAA singlet around 2 0 ppm could be weak compared to the cholines around 3 2 ppm or the creatines or the lipid or macromolecule signals around 1 3 or 0 9 ppm So this is not only for tumors See PLOTs 6 amp 7 in HOME 1cmodel doc figures pdf for examples and Sec 9 1 for instructions e Muscle Input SPTYPE muscle 5 for muscle spectra See PLOT 8 for an example and Sec 9 2 for important limitations and instructions e Lipids Breast Liver etc For spectra with mainly lipids and water and possibly choline Examples are in PLOTs 9 11 To use these you must read Sec 9 3 19 20 CHAPTER 3 ESSENTIAL GUIDE e MEGA PRESS Input SPTYPE mega press 3 for quantifying GABA 7 using MEGA PRESS To use this you must read Sec 9 4 e CSF NAA cholines amp Cr all weak Input SPTYPE csf in the unusual case that you have mainly cerebrospinal fluid CSF with the three usual landmarks NAA cholines and creatines weaker than Lac amp Glc PLOT 13 shows an example 3 2 Chemical Shift Imaging For GE CSI data sets use the Research Version of SAGE which now has Mary McLean s LCModel interface 9 For this SAGE LCModel interface you must use Sec
128. den Control Parameter FWHMST lt 0 and it must be positive Hints Do not input this FATAL A starting value for a y is not positive and they must be Hints Check that your Control Parameters DEEXT2 ALEXT2 are positive leave the Hidden Control Parameter FWHMST at its default FATAL Same as SETUP 5 SOLVE sets up and solves the quadratic programming problem at each iteration of the nonlinear least squares analysis of 1 Eq 4 1 2 3 4 5 6 7 ILLOGICAL Meaningless arguments in call to SOLVE FATAL One of the o e lt 0 and they must be positive Hints These values are given in the table of Basis Spectra in the Detailed Output Check that your Control Parameters DESDSH ALSDSH are positive FATAL You have set the Hidden Control Parameter SDMSHF nonpositive and it must be positive Do not input it FATAL Same as SOLVE 2 FATAL One of the o y lt 0 and they must be positive Hints These values are given in the table of Basis Spectra in the Detailed Output Check that your Control Parameters DESDT2 ALSDT2 are positive FATAL Same as RFALSI 3 FATAL One of the following Hidden Control Parameters has been changed to 170 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS a nonpositive value DFLDMQ DDGZMQ DDGPMQ RSHFMQ RPOWMQ RSDSMQ RRT2MQ NSIDMN Do not input these 8 ERROR Somehow there are as many free parameters as data points or the fit to the data is perfect Hints
129. dor Specific Guidelines e o 24 Sil Bruker s std aa Ba e a dle da 24 Sia ME A A AAA A a ek oe SS 24 A A be Ae ho en Aa ie J 25 3 7 4 Marconi Picker de ta a le hoe A 25 3 7 5 Bruker Philips Siemens Toshiba Varian Microsoft 25 JLO UA gh he A a i ae et E al oes 25 IT T SIEMENS Te re 504 rt Gah Abel ode ae Vee oA foe on 26 A o Rte Scat Ade Aan ad OF tithe Peet BE 26 3 7 9 Varian Agilent oe Sd Be A hd Moe eg Mek A aac 27 3 7 10 Other Vendors o sos a sor s ya soe e a o 27 4 Installation and Test Runs 28 Al Postacript Output maria ed ee se da a a 28 4 2 Test Runs Using LCMgui 2 0 2 02 02 0004 28 4 2 1 Display on Your Screen 2 2 0 0004 28 4 2 2 PostScript Printer iia Gd Ate ela eee Da Pe 29 4 2 3 No Printer or Display o e e 30 ALA Further Tests sica tina doe eos ita cd eg eae Ree Bons 30 4 2 5 More Complete Test of LCMgui with a False License 30 4 2 6 Starting Over s caras a a a 31 4 3 Test Runs without LCMgui e e 31 4 4 Benchmark Vimings senienas morta hie nl te 32 CONTENTS 5 Running LCModel without LCMgui Basic Input 5 1 Conventions 5 2 5 3 5 1 1 File extensions 5 1 2 Control Parameter Conventions RAM Elle us Aten ie os le A et ee eee 5 2 1 Namelist SEQPAR 5 2 2 Namelist NMID 5 2 3 Time Domain Data CONTROL
130. e SDAT file with LCMgui In Fig 6 4 switch off Do eddy current correction 3 7 7 Siemens Quite a variety of files single voxel and 2D amp 3D CSI are automatically recognized and handled by LCMgui With the syngo PC platform you must transfer the files to your Linux or Unix workstation With FTP transfer with binary mode not text and not ASCIT mode Use the syngo rda files They are generated in the spectroscopy card via options gt export raw data Most syngo DICOM ima files of course with time data not image data can also be used They are generated by the command transfer gt export to off line in the patient browser Future DICOM files may not work LCMgui can also handle syngo raw binary files with no header but the above rda and ima types are much more complete and convenient to use Vision single voxel raw files from the Sun Unix platform are also handled automatically with LCMgui With single voxel spectra you should acquire an unsuppressed water ref erence spectrum from the same voxel You then select Do eddy current correction in LCMgui Sec 6 3 3 You will then be prompted to select the unsuppressed water reference file Estimates of absolute concentrations using calibration phantoms Sec 10 1 is only convenient with Vision raw files and only when the transmit coil is also the receive coil In other cases a more elaborate calibration is nec
131. e Step 6 above Enter a unique metabolite name In this way your ECC data in the format of RAW files for direct input to MakeBasis go into the files usr myname press25 ecc where the are your names for the metabolites e g naa gaba etc e Step 9 above Click Run LCModel e Step 12 above Click Next Analysis If you have no more ECCs to do click Exit LCMgui Then click Exit amp Forget all of the strange settings above will be forgotten and your next LCMgui session will offer the usual defaults for an LCModel analysis If you click Exit amp Save by mistake in Step 12 you can quite easily undo the above changes in the next LCMgui session To undo Step 4 you usually save TABLE and PS files To undo Step 5 you usually simply click on Change Configuration and then on Restore Defaults 8 4 2 ECC without LCMgui If you convert your data to LCModel RAW files Sec 5 2 you can use the program KECC to apply the Klose correction 10 as explained in Sec 9 6 An example is HOME 1cmode1 bin kecc lt gaba keccin The output file gaba ecc contains the eddy current corrected time domain data in the format of a RAW file You input the ecc file without any changes in place of the RAW file in Sec 8 6 3 The input file gaba keccin is the Namelist NMKECC FILRAW gaba RAW FILH20 gaba h20 FILOUT gaba ecc NUNFIL END where gaba RA
132. e corrections will be made info An additional search was made for phase corrections but the original solution was found to be best UPDPRI Update priors Updates prior probabilities for phases amp shift during a multi voxel analysis WSCALE 1 FATAL Same as MAIN 1 Water Scaling Does Water Scaling Errors here mean that if you are attempt ing Water Scaling DOWS T then only the concentration ratios of the metabolites in the BASIS file will be valid If you are simulating metabolite Basis Spectra then the concentrations of all of the simulated metabolites would have to be mul tiplied by an unknown scale factor to be consistent with the concentrations of the BASIS metabolites All WSCALE diagnostics are of type ERROR 1 2 The integral of the unsuppressed water reference signal is not positive or could not be computed because of AREAWA 1 4 Hints Your water reference must be without water suppression Check and or display this data in FILH20 One of the following was not input positive ATTH20 WCONC 174 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS ZEROVX Zero voxels Opens and checks the RAW file In multi voxel analyses it skips over voxels whose data are all zeros which can occur in Philips SDAT files if you do not follow the steps in Sec 3 7 6 1 FATAL Namelist NMID in your RAW file contains no or a blank FMTDAT format specification The data cannot be read 2 FATA
133. e file HOME 1cmodel gelx preprocessors set 3t bandwidth as instructed in the comments there Then select this Preprocessor in Fig 7 15 use unsuppressed Cannot be used with multi channel phased array data It will cause the unsuppressed water reference frames to be analyzed instead of the usual water suppressed frames You acquire many water reference frames and only a small number of water suppressed frames there still must be at least one although it is not used This is used instead of Water Scaling to get a direct and unified analysis of a single spectrum including the water signal for example for bone marrow Sec 9 3 7 4 6 Changing the License amp the Print Command These settings were made during installation You can change them as follows 7 5 FINE POINTS 75 e In the main Control Parameters window click on Advanced Settings to get the menu in Fig 7 6 e Click on Change LCMgui Settings This produces a window with entries containing the current OWNER and KEY license settings as in Fig 7 1 and the print or display command that LCMgui is currently using for outputting the LCModel PostScript file With Linux there are no OWNER and KEY entries e Edit any of these entries that you want to change and click on OK or click Cancel Changes to cancel your changes and leave everything as it was One common change is to delete the entire print command leaving the entry blank which
134. e of the spectrum With LCMgui you can use Preview Data Sec 7 4 1 Otherwise input DOREFS 2 F to LCModel as in PLOT 5 in HOME 1cmodel doc figures pdf From the plot estimate the referencing shift You can then directly impose this starting shift with PPMSHF PPMSHF REAL shift in ppm LCModel will start looking for the referencing shift at PPMSHF remember the LCModel convention that a positive shift shifts the data toward higher ppm LCModel will start its search only at this one shift the usual elaborate grid search is suppressed Note that the disaster in PLOT 17 was provoked by using a grossly incorrect PPMSHF Before you go to the trouble of plotting the absolute value you might simply try PPMSHF 0 this is often close enough except for phantoms at room temperature which usually need a PPMSHF 0 18 If the magnitude of PPMSHF exceeds the total ppm range of your Analysis Window obviously a senseless value then PPMSHF is not used and the usual grid search is performed Default PPMSHF 1 E37 i e PPMSHF is not used An example Suppose the Cho peak which should be at 3 2 ppm is fit as Cr at 3 03 ppm We need to shift Cho from 3 03 to 3 2 i e by 0 17 However suppose the table MISCELLANEOUS OUTPUT shows Data shift 0 050 i e the observed spectrum has already been shifted by 0 050 Therefore the total shift of the original spectrum must be PPMSHF 0 17 0 05 0 12 136 CHAPTER 11 FINE POINTS 11
135. e spectrum warning The first order phase correction is below DGPPMN SDDEGP or above DGPPMX SDDEGP Hints If you are expecting such extreme first order phase corrections then adjust DGPPMN DGPPMX DEGPPM SDDEGP accordingly Otherwise the hints of FINOUT 2 apply info The lineshape is not very smooth Hints These lineshapes occur more often at low resolution or long echo times where there are relatively few peaks in the spectrum This generally does not seriously affect the concentrations since they are obtained by ef fectively integrating over the lineshapes and the oscillations tend to cancel out However the FWHM listed in the One Page Output but not used for anything else may be wrong warning The zero order phase correction is more than 4 standard devi ations SDDEGZ from its expectation value DEGZER The phasing may be poor and you may have to reject this analysis Hints Large artifacts or errors in the data can in rare cases cause this Much more common is that the prior information that you are imposing is wrong Check for a wrong value of DEGZER default is 0 0 or too small an SDDEGZ This prior information often assumes that an eddy current correc tion was done Did you forget this or did you use the wrong water reference file With CSI data do not input DEGZER or SDDEGZ Let the Bayesian learn ing in LCModel set these SDDEGZ is usually very large with CSI anyway Occasional occurrences of this diagno
136. e stage shown in Fig 7 5 If you are satisfied with the structure shown in the field next to Con structed Save Directory then click the Set Configuration button This special dot star combination at the end of the Archive Directory causes LCMgui to make a file extension In this example the PS file will be SP sp 1lcmodel saved vol_2401_148 ps Without the dot star at the end the file would be SP sp 1cmodel saved vol_2401_148 ps i e a slash would replace the dot and the extension would become a file 7 3 4 View Edit Control Parameters LCMgui gives you the full capability of comfortably setting all Control Parameters in this manual You can also save these settings for optional use as defaults later 64 CHAPTER 7 LCMGUI REFERENCE MANUAL zj Conf igure Save Director FES Here you configure the structure of the default directory where your LCModel results will be saved It will then be offered in the Control Parameters window The configuration is constructed by one or more successive elements that you select below Clicking on the green Append buttons below will append elements most that will change with each spectrum yielding unique directories You can also enter fixed elements in the green field below and click on Append Entry This can also be a single to separate subdirectory names The yellow field shows the Save Directory that you have constructed so far Click on Start Over to clear the
137. e that Cho is undetectable S N is output in the MISCELLANEOUS OUTPUT table Sec 2 4 9 4 MEGA PRESS for GABA This is for the MEGA PRESS specified in 7 PLOT 12 in HOME 1cmode1 doc figures pdf shows an example 9 4 1 Input 9 4 1 1 Necessary input You select the final difference spectrum Your input is simply SPTYPE mega press 3 PPMST 4 2 112 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION PPMEND 1 95 LCMgui users use Fig 7 7 in Sec 7 3 4 The above PPMEND cuts off dangerous artifacts that usually occur between 1 95 and 1 2 ppm Only if they are nearly absent as in PLOT 12 can you use PPMEND 0 2 With PPMEND 1 95 part of the NAA peak is truncated However LCModel uses the complete spectra not peak integration and the incomplete NAA peak can still provide an accurate estimate If you also want to quantify the macromolecule peak around 0 9 ppm MMO9 e g to get a rough idea of the contribution of macromolecules to the peak around 3 ppm You can cut out most artifacts by inputting Sec 11 11 PPMEND 0 2 PPMGAP 1 1 PPMGAP 2 1 1 95 12 9 4 1 2 Additional necessary input with Philips files With Philips MEGA PRESS files be sure to also input IAVERG 3 since the data set is usually a series of alternating on and off spectra that are summed using IAVERG 3 Sec 11 9 2 If you switch from MEGA PRESS to other data be sure to delete TAVERG in Fig 7 7 If you are doing Water
138. ea the ratio of the metabolite resonance area to the unsuppressed water resonance area If you want to convert this to Concmet the concentration of the metabolite groups in mM for example Concmet Ratioarea X 2 N1Hmet X ATTH20 attmet X WCONC N1Hmez is the number of equivalent protons contributing to the resonance e g NlHmet 2 for mM of CH2 groups not mM of lipid molecules This becomes meaningless for Lip16 Lip09 Lip13 which has CH3 Lip09 and CH2 Lip13 Lip16 groups therefore resonance areas are more convenient ATTH20 amp attmet are the factors lt 1 by which the water and metabolite resonance areas are attenuated by relaxation With Case 2 above ATTH20 might also have to correct for the decreased total gain of the unsuppressed water reference spectrum compared to the water suppressed spectrum see Sec 9 2 2 4 WCONC is the water concentration in mM in the voxel and this is usually difficult to guess 9 3 3 Estimating only choline Particularly in breast fat suppression can distort the lipid signals around 1 3 ppm and possibly 2 0 ppm as well making their quantitation meaningless Usually water suppression is also necessary The only interesting undistorted signal is then Cho around 3 2 ppm You can then estimate the Cho by following Sec 9 3 1 1 except Step 1 where you input the following SPTYPE only cho 2 PPMST 3 8 PPMEND 2 7 PPMST and PPMEND are also very important 9 4 MEG
139. each Basis Spectrum The first time that you do an ECC you have to change settings in LCMgui to execute the program KECC rather than LCModel After you have done all your ECCs you exit LCMgui with Exit amp Forget and all these strange settings are forgotten and you have your usual default settings for LCModel in the next session 1 2 al wa OS 00 2 10 11 Select the file with your water suppressed data In the Control Parameters window change the two values next to An alyzing spectrum from to 4 0 ppm and 0 5 ppm This only determines the range of the plots of the data If necessary click the Do eddy current correction button to make it blue In the Save File Types menu in Fig 7 3 select only the TABLE to be saved No TABLE file will be produced and you will get an error message in your Unix window However it is necessary that something be selected for saving so that the Save Directory is made available for output of the ECC data To specify a directory to receive your ECC data click on Reconfigure to produce the window in Fig 7 4 For example you could enter home myname press25 in the field next to Enter a fixed element Put a slash at the end Then click Append Entry and then Set Con figuration This is now shown in the green field next to Save File types to directory Append a unique name for the metabolite in you
140. each other To make use of this use the following protocol Make one single huge Stan dard Solvent of 1 mM DSS TSPS 72 mM KHPO 28 mM KHPO 200 mM Na formate not Sigma F4166 or F6502 1 g L NaNz At fields of 4T and above you may have to reduce the ionic strength by reducing the phosphate concentrations At fields of 9 4T and above reduce DSS to 0 5 mM Dissolve each of your model metabolites in this Standard Solvent In this way the formate concentration is exactly the same in all phantoms and MakeBasis can use the resonance area of the formate singlet around 8 44 ppm to scale the spectra consistently with each other which is absolutely essential The formate and DSS singlets are also used by MakeBasis for automatic phasing If a metabolite such as the CdCly adduct of GPC or the Ca salt of PCh cannot be dissolved in the Standard Solvent due to the phosphate be sure to dissolve it in 200 mM formate and about 2 mM DSS so that the Auto Scaling and Auto Phasing can still be done The formate concentration need not be exactly the same as in the Standard Solvent as long as you accurately know the formate concentrations in each pH control is essential particularly for NAA amp Glu otherwise some of their resonances shift significantly and the entire LCModel analysis can be ruined For several metabolites particularly NAA Lac Glu you will have to adjust the pH back to 7 2 If you titrate carefully with concentrated
141. eady been caught with MYCONT 1 or MYCONT 2 is the main program 1 FATAL Input or output error with the internal file filcsi_sav_1 or filcsi_sav_2 Hints Correct your changes to the Execution Script standard or nice Do not input the hidden Control Parameters filcsi_sav_1 filcsi_sav_2 1csi_sav_1 or 1csi_sav_2 2 FATAL Error trying to open FILCSV Hints Check permissions 3 FATAL Error trying to read FILCSV Hints Check permissions organizes the production of the PostScript file for the One Page Output 1 FATAL At least one of the Control Parameters PAGEHT or PAGEWD is not positive No One Page Output can be produced Hints Correct your input of PAGEHT or PAGEWD 2 FATAL Error trying to open FILPS No One Page Output can be produced Hints Check permissions Merge left merges Baseline segments to be continuous across gaps 1 ILLOGICAL Programming error Merge right merges Baseline segments to be continuous across gaps 1 ILLOGICAL Programming error inputs the Basis Spectra from the BASIS file and simulates additional Basis Spec tra 1 FATAL You are using a basis set that requires LCModel version 6 3 1 or higher Hints Update to the latest LCModel Sec 7 2 6 2 warning More than 90 of the Basis bandwidth BW is needed to cover the Analysis Window There will probably be distortions at the edges FATAL The Basis BW is narrower than the Analysis Window Hints Use a Basis with a larg
142. eak would cause an extreme shift of SHIFMN 1 or SHIFMX 1 Thus the residual water peak after water suppression is very far from 4 65 ppm Hints This is often caused by a very distorted or weak residual water peak If this diagnostic consistently occurs plot the absolute value of the spectrum including the water peak with Preview Data in LCMgui Sec 7 4 1 If your water peaks are consistently very weak and distorted then do not in put DOREFS 1 T leave it at its default of FALSE However except for wasted computer time this diagnostic should imply no problems unless the landmarks specified by the second set of 6 functions are also very weak If a disaster like PLOT 17 in HOME 1cmodel doc figures pdf occurs Sec 11 3 can help A second possibility is that this extreme shift really is necessary You can also see this from the Absolute Value Plot If this is so then input PPMCEN or adjust SHIFMN 1 or SHIFMX 1 warning The CCF with the second set of 6 functions specified by PPMREF 2 would cause an extreme shift of SHIFMX 2 Hints This is usually caused by a spectrum with little structure or with a huge lipid peak This could cause a gross referencing error and something like PLOT 17 in HOME 1cmodel doc figures pdf If this occurs try SPTYPE tumor only with brain or plot the Absolute Value Plot of the spectrum with Preview Data in LCMgui Sec 7 4 1 and use PPMSHF A second possibility is that th
143. eaningless argument in call to ERRMES FATAL Stopping after MERMES 2000 diagnostics FINOUT does the computations and error analyses for the output at the end of the analysis 1 10 11 ERROR An analysis to rephase the data for plotting failed The plot will not be of the real part of the data The concentrations are not affected Hints Never Happened ERROR Giving up after maximum iterations in trying to rephase the data for plotting The results are as in FINOUT 1 Hints This is a sure sign that your data are so poor that the phase correc tions are very uncertain you should probably reject the data FATAL An analysis to get error estimates failed Hints Never Hap pened but if it does then your data should be rejected anyway FATAL Same as FINOUT 3 FATAL Same as FINOUT 3 warning The fit to the data was perfect and the S N therefore cannot be calculated and is being set to 1 0 Hints Never Happened but if it does it is probably because all data points are zero info The Baseline required a great deal of freedom and may have a wild shape Hints This may be due to shimming problems or to susceptibility or other artifacts or to very sharp residual water or lipid peaks If the problem is near the edge of the spectrum then you should eliminate this using PPMST or PPMEND If the spectrum being analyzed is in vitro then try inputting VITRO T If the problem persists then you should probably reject th
144. ed any Control Parameters e g activated the Do water scaling button a window offering you the chance to save these changes for possible future use as defaults appears This can be very useful in saving you from having to remember and enter these changes in the future This is described in Sec 7 3 5 6 5 Where is the Output Normally LCMgui automatically outputs the One Page Output with a print or display command that was specified during installation in Sec 4 2 2 If you cannot find your output Sec 7 4 6 specifies how you can check and correct this print or display command There are several other useful files that can be output and archived by LCModel Clicking on Save File types in Fig 6 4 produces a menu where you can select which types of output files will be archived In addition to the PS file with the One Page Output the TABLE file with tables summarizing the results should be selected The green field following Save File types to directory which you can edit shows the directory where they will be archived See Sec 7 3 3 which also specifies how you can change the structure of this Archive Directory tree By default the filename is simply the type of output e g ps is the PS file and table is the TABLE file By default for multi voxel CSI data sets they are made unique as in the following example The Slice 2 Row 3 Column 4 PS filename is s12_3 4 ps The INPUT CHANGES table in your One
145. ee 11 1 2 Normal Use of LCModel 0 020020 000 12 1 3 Conventions amp Notation 0 0 a 12 IA ai 5 2 seh ere Seo ee Rt a Ban he SE Set dn es ee 13 1 4 Acknowledgments 002002 pee ee eee 13 2 One Page Output 15 2 1 Concentration Table o 020200005022 15 ZLI Upper Bart emos dato dl da a ra amp 16 ZUZ MO WEB Parts ge A A Os ees 16 2 1 3 Prior Ratio Probabilities 17 2 2 SPloty us fei aang Bad a ae on aah Be EE RUN ee 2 ot 17 2 37 Diagnostics Lita Mee phe co Bday ek i Ay oe Hog Ae ds 17 2 4 Miscellaneous Output o 18 2 5 Input Changes rociado A a ee ae da a E 18 3 Essential Guide 19 3 1 Special Spectra a lt 4 ek a A ae 19 3 2 Chemical Shift Imaging e o 20 3 3 Acquisition Parameters e 20 3 4 Analysis Window e 21 do PPMEND gregi a e aa di a Bead Sek 21 A PPMST wy ve ay cena ei docs dle A de Sey A os Ge A ted 21 3 5 Criteria for Rejecting Analyses e 000004 22 4 CONTENTS 3 5 1 Non Random Residuals o a 22 352 Wild Baseline piro ara a r en oe eS ie Pen a rada 22 3 6 Causes amp Remedies for Failures o 23 3 6 1 Artifacts s iia Si ee poet ead De de a a da 23 3 6 2 Referencing 0 0202 eee eee eee 23 316 3 Phasing 40 4 00a eben GF eek Gets Be ia cea 23 3 6 4 Smoothing 2 2 4 4 6 pak a he a a a a e a 23 3 7 Ven
146. eighting and must be in the same CSI format You must also activate Water Scaling by inputting DOWS T IAVERG 3 to analyze an unweighted average of the spectra i e the spectra are simply summed and divided by the number of spectra This is not too bad if the scalings and noise levels in the spectra are not extremely different IAVERG 31 to do an unweighted average of only the odd spectra e g for Philips MEGA PRESS Sec 9 4 1 2 IAVERG 32 to do an unweighted average of only the even spectra e g for Philips MEGA PRESS Sec 9 4 1 2 Philips often sets the number of rows columns amp slices in the SPAR file automatically Check the multi voxel panel in Fig 6 4 and set them to correspond to the number of spectra if necessary You can also view the spectra with the Preview Data button in Fig 6 4 You can get a squarer preview by setting the number of rows and columns approximately equal e g going from 1 x 40 to 5 x 8 This adjustment does not affect the analysis 11 9 3 Varying Phases or Referencing Shifts Even in a series of scans of the same voxel motion especially in the pelvic region can cause the phases to vary These motion or other artifacts can sometimes be reduced by partitioning subsets of scans into separate files phasing and referencing each with LCModel averaging and analyzing the average with LCModel You can do this with the following labor intensive steps You analyze each file separately The
147. els Amsterdam Jan Weis Upp sala Noriaki Hattori Osaka and R Mark Wellard Brisbane for providing model spectra and Xiangling Mao Gottingen for information on Siemens data Section 8 2 is mainly based on the results of Thomas Michaelis and Petra Pouwels 14 CHAPTER 1 PREFACE amp OVERVIEW The code generating the PostScript output files is based on advice and routines kindly provided by Christian Labadie Dept of Chemistry SUNY Stony Brook I am grateful to Gunther Helms Tiibingen Else R Danielsen Copenhagen Markus Wick Ettlingen Uwe Seeger Tiibingen and Jens Frahm Gottingen for informa tion and suggestions on scaling for absolute concentrations I thank Uwe Seeger also for making 3 available before publication and for helpful discussions on including lipids and macromolecules in the analysis I thank Brad Beyenhof for advice on for warding X windows I thank Philipp B hm Sturm Berlin for much information and advice on Bruker data LCMgui was written with the free software packages Tcl Tk http www tcl tk Tix Toi K Lam http tixlibrary sourceforge net and Plus Patches Jan Nijtmans http members1 chello nl j nijtmans plus htm1 Chapter 2 One Page Output The results of an LC Model analysis are summarized on the so called One Page Output now usually two pages contained in a PostScript file Print out HOME lcmodel doc figures pdf PLOTs 1 amp 2 there show the One Page Output from
148. end me two or three dump files with high quality PRESS spectra Siemens data often have a big However this is usually quite consistent i e also with a small SDDEGP as with the Marconi Picker values which were calculated from their precisely known timing Section 9 7 1 shows how you can experimentally estimate DEGPPM and SDDEGP 5 3 6 PostScript Output Section 9 9 3 gives many possibilities to tailor the format of the One Page Output to your wishes even for preparing slides or figures for publication In any case you should input the following Control Parameter PGNORM CHARACTER 2 page norm US sets the page dimensions of the One Page Output to 8 5 x 11 0 inches A4 or EU sets the dimensions to 21 0 x 29 7 cm The default is PGNORM and makes a compromise to fit in both page formats Usually there is not enough room for the complete tables to the right of the plots in the One Page Output Then by default a second page with the complete tables and no plot is produced Therefore you should use a duplex printer if possible so that both pages are on one sheet IPAGE2 INTEGER IPAGE2 0 to suppress printing the second page e g if you do not have a duplex printer thereby losing information IPAGE2 1 the default to output a second page PLOT 2 in 42 CHAPTER 5 RUNNING LCMODEL WITHOUT LCMGUI BASIC INPUT HOME 1cmode1 doc figures pdf with the complete tables if the
149. er BW or narrow the Analysis Window No longer used No longer used 5 FATAL The number of Basis Spectra to be used exceeds 60 Hints If you really want to use more then I can send a new version Ae 0 162 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS 6 10 11 12 13 14 15 16 17 18 19 FATAL At least one of the following input variables in a Namelist BASIS in your BASIS file is not positive TRAMP VOLUME CONC The last line in the table of Basis Spectra in the Detailed Output gives the Metabolite Name corresponding to the Basis Spectrum involved FATAL All of the metabolites in your Basis Set have been omitted from the analysis there are none left Hints a This may be due to your BASIS file having almost no metabo lites b These could also have been omitted by your input of NOMIT and CHOMIT or of BASCAL and CHBCAL or of NCALIB and CHCALI c They could have been automatically omitted if your PPMST and PPMEND produce a very narrow Analysis Window d The Preliminary Analysis uses only the metabolites named in CHUSE1 NAA Cr or Cre GPC or PCh or Cho Glu Ins by default So at least one of these must be in the analysis However in order for the Preliminary Analysis to work properly all of the major landmarks in the spectrum being analyzed should be included i e you need at the very least NAA Cr Cho and at short TE Ins except possibly with a phantom If you are analyz
150. er examples and specifications on preparing your own my bin2raw script or executable program Par ticularly philips bin2raw contains arbitrary assumptions e g on filenames which you can easily modify to your needs 7 4 FURTHER USEFUL SETTINGS 73 l r Select Preprocessor XA Below is a list of Preprocessor programs You can select one of these programs to further modify the Control Parameters just before LCModel runs Some possibilities such as using your unsuppressed water reference for absolute concentrations are discussed in the LCModel Manual The Preprocessor program if any that has been used so far in this session is shown in green right below Selected File You can select another Preprocessor by clicking on its name below Click on Do Not use any file below to use no Preprocessor at all Click on Use Selected File below when you are satisfied with the green file No entry in the green field means that no Preprocessor will be used You can later check and edit the Control Parameters resulting from your selection in the View Edit Control Parameters window in the Advanced Settings menu The Preprocessor files are inthe directory SP sp lcmodel gelx preprocessors Selected File ges select frames sset 3t bandwidth Use Selected File Do Not use any file Figure 7 15 Here you can view and or change the Preprocessor that will be used by LCMgui Even after you make other my b
151. error message in your Unix Linux window often with a Standard Diagnostic Sec 12 2 Often one of the most important questions in troubleshooting is whether the data have been correctly converted or read in Check this by plotting the input data with the Preview Data button in LCMgui Sec 7 4 1 With LCMgui all run time messages including error messages are logged in the file runtime messages in the temporary directory for this analysis Sec 7 5 2 2 With out LCMgui they appear in the Linux Unix window Linux messages about Skip ping namelist are normal 12 1 1 I O Errors This is mainly for non LCMgui users Input Output errors are the most common errors especially during the first few runs Some systems give good diagnostics in 153 154 CHAPTER 12 DIAGNOSTICS AND TROUBLESHOOTING HINTS runtime messages often showing exactly where the I O error occurred End of File during Read The most common cause is a missing or incorrect Namelist name The system reads the file until it finds the Namelist or an end of file Violating the Namelist rules for spaces in the first column or characters in the second column can cause the system to miss your Namelist If this occurs with a RAW file also check that the format of your time domain data agrees with FMTDAT in Namelist NMID and that there are at least NUNFIL complex pairs in the RAW file Namelist Errors These usually involve simple rule violations or
152. font size 76 FORECC 104 frames selecting GE 74 FTP transfer modes 25 FWHM 18 FWHMBA 92 GABA 111 INDEX Gaps in ppm axis inserting 147 GE data 24 126 145 bandwidth at 3T 74 phase corrections 74 selecting frames 74 GE SAGE LCModel interface 57 GLOSCA 88 glycine 79 Grid Point 89 gui defaults LCMgui 75 H20 file LCMgui 60 H20 file 115 Hidden Control Parameters 151 HIFMM 142 Hitachi data 25 HWDPHA 93 HWDSCA 93 HZPPPM 35 38 HZREF 136 IAVERG 146 ICOLEN 39 ICOLSK 139 ICOLST 39 ID in vitro 91 in vivo 36 IDBAST 93 IETCOU 122 ILLOGICAL 155 IMCL 103 IN file MakeBasis 92 PlotRaw 87 info 154 institutional concentration units 128 interactive processing LCMgui 51 71 internal water scaling 128 IPAGE2 41 IROWEN 39 IROWSK 139 IROWST 39 ISDBOL 119 ISHIFT 98 ISLICE 39 KEY 37 181 LANDSC 121 LBASIS 114 LCMODL Namelist 37 LCONTR 114 LCOORD 43 LCORAW 147 LCSV 42 lesion amp tumor spectra 103 license changing in LCMgui 74 installing in LCMgui 56 lipid spectra 102 lipids simulating 140 liver spectra 102 LPRINT for LCModel 150 for PlotRaw 89 LPS 114 LRAW 114 LTABLE 42 macromolecules simulating 140 magnitude spectra 151 MAKEBA diagnostics 96 MakeBasis diagnostics 96 MakeBasis program 90 Marconi data 25 41 58 62 95 MEGA PRESS 96 111 merging multi voxel PS files 139 METABO 94 Metabolite Name 94 me
153. for in vivo data However Namelist SEQPAR is not used and the values of the following Control Parameters in Namelist NMID are more important with the Basis Spectra ID You should give a unique name to each Basis Spectrum at least if your file names are not unique e g if you call all NAA data files simply naa RAW However it is much better to also use unique filenames VOLUME is not important if you are using Auto Scaling Otherwise it should equal the voxel size say in mL so that the Basis Set will be consistently scaled even if you use different voxel sizes although a fixed voxel size for the Basis Set is recommended 92 CHAPTER 8 MAKING THE BASIS SET TRAMP is not important if you are using Auto Scaling A reasonable value is 1 0 If you do not use Auto Scaling then you must be especially careful that all the spectra for a Basis Set are acquired under identical conditions including coil loading so that the effective gain is the same for all in this case a TRAMP 1 0 could also be used Otherwise you must set TRAMP according to the methods mentioned in Sec 10 1 1 For MakeBasis you must explicitly input the complete Namelist NMID except that only Bruker Philips amp Toshiba users need to input BRUKER and only Bruker users may need to to input SEQACQ Immediately following this Namelist are the raw time domain data see Sec 5 2 3 They are for the metabolite Basis Spectrum possibly corrected for eddy curren
154. go down to 1 0 ppm At 2 9T you need a bandwidth greater than 1200 Hz you need at least 1800 Hz Input SPTYPE lipid 8 or SPTYPE liver 11 or SPTYPE breast 8 or SPTYPE only cho 2 PPMST is very important It is specified in Secs 9 3 1 2 amp 9 3 1 1 Often you want to estimate ratios of metabolite resonance areas to water You then need a spectrum without water suppression and there are two possibilities 9 3 1 1 Two spectra Separate water reference spectrum You acquire a water suppressed spectrum and an additional unsuppressed water refer ence spectrum from the same voxel immediately before or after the water suppressed spectrum GE Probe Picker and Toshiba do this automatically 1 Normally you input PPMST 3 6 when SPTYPE lipid 8 or PPMST 3 8 when SPTYPE breast 8 or PPMST 4 0 when SPTYPE liver 11 If you also want to estimate the lipids around 5 3 ppm you instead input PPMST 8 0 For SPTYPE only cho 2 see Sec 9 3 3 2 Do Water Scaling using Sec 9 2 2 4 3 Eddy current correction is automatically suppressed Sec 9 2 2 3 4 If you have prior knowledge about the phase corrections be sure to input this with SDDEGZ SDDEGP DEGZER DEGPPM Sec 9 7 This is automat ically done with GE Probe and Picker data If you cannot do Water Scaling Step 2 above you must use ratios of the metabo lites Sec 9 3 2 108 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATI
155. gt 4 Hints Do not input NBCKMN 25 FATAL You must have the Hidden Control Parameter DKNTMN 1 or DKNTMN 2 gt 0 Hints Do not input DKNTMN MYDATA reads and scales the data in the RAW and H20 files 1 FATAL Namelist NMID in your RAW file contains no or a blank FMTDAT format specification The data cannot be read 2 FATAL At least one of the following input variables in Namelist NMID in your RAW file is not input positive TRAMP VOLUME or your Control Parameter FCALIB is not positive 3 FATAL Namelist NMID in your H20 file contains no or a blank FMTDAT format specification The data cannot be read 4 warning HZPPPM in the CONTROL file differs from that in the RAW file by more than 5 The HZPPPM from the CONTROL file will be used This could mean poor modeling of coupled resonances if it is wrong Hints Check the values in both files 5 ERROR No H20 file could be read although DOWS Tor DOECC T Water Scaling and eddy current correction will not be done 6 Not used 7 FATAL Error trying to read Namelist NMID in FILRAW Hints See Sec 12 1 1 8 FATAL Error trying to read time data in FILRAW Hints In FILRAW check that the time data correspond to FMTDAT Does the file size correspond to your input NUNFIL or with multi voxel data with NDSLIC NDROWS NDCOLS 9 FATAL Error trying to read Namelist NMID in FILH20 Hints See Sec 12 1 1 10 FATAL Error trying to read time data in FILH20
156. hat you received with your license including capitalization and spacing You do not need to enclose the string in single quotes License OWNER Blomedizinische NMR Forschungs GmbH Goettingen In the green field enter the KEY 9 or fewer digits that you received with your license If you received more than one KEY then enter them all each separated by a space Click OK when you have entered the information above You are then given the option to select a data file to be analyzed OK Abort LCMgui Figure 7 1 The Install License Window with Sun SGI or DEC Compaq example of how you can separate field strengths Suppose that you have 1 5T amp 3T data e mv your 1 5T Basis Sets into the directory HOME 1cmodel basis sets e For your 3T Basis Sets mkdir HOME 1cmodel basis sets 3t and mv the 3T BASIS files into this directory e The first time that you analyze 3T data click the Change BASIS button in Fig 7 2 and then select the appropriate 3T basis file from the 3t directory e When you exit from this first 3T session click New Profile in Fig 6 6 enter 3T in Fig 7 11 and then click Save e At the start of future LCMgui sessions Fig 7 12 allows you to select 1 for your 1 5T User Profile or 3T for your 3T User Profile This way you keep your 1 5T and 3T Basis Sets archives of results default settings etc separate e When you exit from LCMgui sessions after the first 3
157. he program PlotRaw It is useful for documentation but it is optional if you leave it out of the Namelist input then a blank ID is output CHARACTER 80 the Fortran FORMAT specification for your raw time domain data which must immediately follow Namelist NMID See also Sec 5 2 3 below This has no default it must be input REAL the voxel size always in the same units e g mL Default VOLUME 1 0 REAL The data are multiplied by the factor TRAMP VOLUME to scale the data consistently with the Basis Set as discussed in Sec 10 1 1 VOLUME and TRAMP do not affect the concentration ratios they only need to be input for absolute concentrations Default TRAMP 1 0 LOGICAL Bruker data Bruker Philips amp Toshiba data must have BRUKER T which causes all 4 programs in the LCModel package to com plex conjugate the data which is necessary with these data types This Control Parameter was named before I had contact with Philips or Toshiba data Default BRUKER F LOGICAL sequential acquisition This is for Bruker users only You do not need to input this in the normal case of simultaneous acquisition where the real and imaginary parts of the complex pair of time domain data are acquired simultaneously If you do not input it then it will be left at its default value of SEQACQ F Only for Bruker sequential acquisition mode default on some Bruker systems prior to the AVANCE series where the imaginar
158. he CHRATO array NNORAT INTEGER is the number of Metabolite Names whose Concentration Ratio Priors will be omitted Default NNORAT 0 NORATO CHARACTER 60 6 The first NNORAT elements contain the Metabolite Names of the priors to be omitted All priors with metnum Sec 11 8 equal to one of these Metabolite Names will be omitted For example with rodents you should input NNORAT 1 NORATO 1 Tau since they have higher Tau than humans At 4 7T and above this is automatically done and the above input is unnecessary 11 9 Coherent Data Averaging This section specifies how you can combine multiple replicate spectra into one spectrum with an increased signal noise and then analyze this Siemens users should use the Siemens utility for combining Siemens phased array data and skip Sec 11 9 There are three cases 11 9 1 GE amp Toshiba Phased Array Single Voxel Data LCMgui recognizes GE amp Toshiba single voxel phased array data It causes LCModel to do a weighted coherent average over the channels and analyze the resultant spectrum So there is nothing that you need to do this is done automatically You may notice an extra Control Parameter IAVERG 1 Do not change this It tells LCModel that this is a set of phased array spectra to be coherently averaged and not a multi voxel set to be analyzed individually Water Scaling is needed for statistical weighting If for some reason you do not want to do Water Scali
159. he DICOM file By default eddy current correc tion ECC has already been done do not do it again i e do not activate the Do eddy current correction button in Fig 6 4 You can activate the Do water scaling button there Sec 6 3 4 e SDAT files must have the extension SDAT or sdat The corresponding SPAR file must be in the same directory with the identical filename except with the extension SPAR or spar respectively e With single voxel SDAT files If you want to do ECC In the Philips post proc page set Spectral correction to yes Then ECC is au tomatically done and you only have to input the file whose name ends in raw_act SDAT Do not do ECC again in LCModel i e do not activate the Do eddy current correction button in Fig 6 4 If you activate the Do water scaling button in Fig 6 4 when asked for your unsuppressed water reference select the file whose name ends in _raw_ref SDAT e With SDAT CSI files Do not do ECC Your menus will now be different 26 CHAPTER 3 ESSENTIAL GUIDE e Prior to acquisition be sure to change the echo acquisition mode to Half Echo from Maximum e You must use the Delayed Reconstruction option on the raw data on a system equipped with reconstruction hardware such as the scanner itself e Set Spectral correction amp Ringing filter to yes e Export your data as SDAT amp SPAR files e Select th
160. he denominator of fscate in Sec 10 2 1 and fscare multiplies the suppressed signal However there are probably big relaxation corrections since Ernst et al 28 found the major water compartment with T2 80 ms This alone would imply ATTH20 0 7 when T E 30 ms since exp TE T2 exp 30 80 0 7 This is partially compensated by your multiplying each concentration out put by LCModel by exp T E T2 where T2 is now that of the metabolite but this is often ignored when T E lt 35 ms Sec 11 2 You can probably ignore T1 relaxation at least for TR gt 4000 ms since T1 is approximately the same for water and the metabolites and the two corrections then partially cancel Default ATTH20 0 7 There is no reason to measure phantoms but if you did you should input ATTH20 1 0 because of the long T 2 of water in vitro 28 REAL the attenuation of the singlet used in the selected Basis Spectrum specified below with WSMET etc The relaxation effects will be smaller than for ATTH20 at least with short TE since the metabolites in vitro usually have longer T2 than water in vivo Also the Basis Spectra should have been acquired with TR gt 6000 ms so you can usually use the Default ATTMET 1 0 You probably will not want to change the following WSMET WSPPM N1HMET CHARACTER 6 Water Scaling metabolite the name of the metabolite whose singlet in the Basis Set will be used for scaling Default WSMET
161. he fine print Particularly important are your settings in the Control Parameters window Fig 7 2 for the Analysis Window Sec 6 3 1 and deciding whether to use eddy current correction Sec 6 3 3 and or Water Scaling Sec 6 3 4 See Sec 3 7 for important guidelines and for a list of the types of data files automatically handled by LCMgui Also note the following 7 3 1 Vendor Specific Guidelines Bruker Marconi Toshiba Varian To avoid writing outside of HOME 1cmodel dur ing testing the initial default Archive Directory is HOME 1cmodel saved Sec tion 7 3 3 3 specifies how to archive in a more convenient place near your original data Bruker With single voxel data the files acqp and method must be in the same directory as your data file fid If the file method is missing you have a very old ParaVision version The instruc tions at the beginning of the text file HOME 1cmodel bruker bin2raw show you how to proceed GE Section 7 4 5 1 describes Preprocessors that are important in the following special cases e when you are not using automated Probe and your data require large phase corrections e when you want to select frames to salvage a spectrum damaged by patient motion during part of the acquisition e when you are not using the standard 5000 Hz bandwidth with 3T data Marconi Picker Only if you are using PRESS You will probably have to select the Pre processor new press Sec 7 4 5 Prior phase infor
162. he ratio multiplied by CONREL but the default is CONREL 1 0 If you want to output absolute concentrations in the third column assuming that say the total creatine concentration was 6 5 then you would simply input CONREL 6 5 LCModel attempts to automatically make reasonable extensions to your NAMREL For example NAMREL Cho will be replaced with a Concentration Sum of all that are present of Cho PCh GPC Similarly the creatines will be extended to include all of Cr Cre PCr 9 9 1 3 Fonts amp Colors ISDBOL INTEGER Integer SD for boldface A line in the Concentration Table will be in boldface if its SD rounded to the nearest integer is less than ISDBOL Default ISDBOL 20 RGBBOL REAL 3 RGB for boldface The boldface line defined by ISDBOL will be in the color defined by RGBBOL as explained below Default RGBBOL 0 0 0 0 0 999 i e blue RATIPM REAL If a ratio with a Concentration Ratio Prior Sec 11 8 is more than RATIPM standard deviations from its expectation value then a or a is inserted before the Metabolite Name in the table if higher if lower The line in the table is bold red italic This is common with the highly variable lipid amp macromolecule 120 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION components Default RATIPM 2 0 RGBRAT REAL 3 RGB for ratio priors The bold italic line defined by RATIPM will be in the color defined by RGBRAT as exp
163. hese identifiers Section 11 4 specifies how to combine PS files into a single file and how skip over a voxel in the rare case that it causes an abort breaking off the multi voxel analysis 5 3 3 Analysis Window You select the Analysis Window the window of frequency domain data that will be analyzed as follows PPMEND REAL ppm end the lower limit of the ppm range i e the right edge of the Analysis Window with the usual convention of the ppm axis decreasing toward the right Defaults See Sec 3 4 1 PPMST REAL ppm start the upper limit of the ppm range i e the left edge of the Analysis Window Default PPMST 4 0 40 CHAPTER 5 RUNNING LCMODEL WITHOUT LCMGUI BASIC INPUT These two Control Parameters are the most important ones for salvaging problem spectra They are probably the only two that the Normal User will also want to occasionally change See Sec 3 4 for guidelines 5 3 4 Eddy Current Correction Philips users can skip Sec 5 3 4 you should have Philips do the eddy current correc tion ECC as specified in Sec 3 7 6 With single voxel brain data eddy current correction 10 is recommended This requires also a reference fid from the same voxel with the same sequence except with no water suppression With a CSI data set all of the voxels in the unsuppressed reference data set must be the same as those in the suppressed data set However ECC is often ineffective with CSI data and is seldo
164. iations in the prior is straining the prior and this Basis Spectrum is probably incompatible with the data A line printer plot of the lineshape coefficients Sn in 1 Eq 1 convoluted with a Gaussian with a FWHM of FWHMBA This gives a rough idea of the final lineshape of singlets g The triangular table of correlation coefficients between the metabolite concentrations This table is useful for seeing strong pairwise correlations e g between NAA and NAAG If the correlation between a pair is consistently more negative than 0 3 then you should also output the Concentration Sum of the pair Sec 9 9 1 1 More complicated correlations among groups of three or more metabolites are not so clear the best reliability indicator is still the SD in the One Page Output which accounts for correlations of all orders and many other things as well To get this output with LCMgui you simply activate the button next to PRINT in Fig 7 3 Sec 7 3 3 1 Otherwise you can use the following FILPRI CHARACTER 255 the name of the text ASCII file that will get the Detailed Output e g pat_777_77 print LPRINT INTEGER LPRINT 0 the default will suppress creation of this file LPRINT 6 will make this file 11 15 Default Control Parameters Most of the default values for the Control Parameters have been listed in this manual In case you want to check these defaults you can do the following First make a file say dum
165. ield The following are the default structures Bruker Marconi Picker Toshiba Varian HOME 1cmodel saved see Sec 7 3 3 3 for a good alternative GE raw P files the standard SAGE data base hierarchy HOME lcmodel saved Site PatientID Study Series GE spectrum G files As with P files except that your Site is replaced with from G file since LCMgui cannot recover the Site from G files Philips HOME 1cmodel saved examination name patient_name scan_date An alternative script where scan_id is also available is available from me Siemens HOME 1cmodel saved lt PatientName gt lt the part of the DataSetId following gt Here Site PatientID Study Series examination name patient name scan_date PatientName amp DataSetId are all taken from header information which can vary with each spectrum Therefore the directories are often unique for each spectrum If you are satisfied with these defaults then you can skip the rest of Sec 7 3 3 You can edit the green field next to to directory but this new directory will only be used during this one analysis 62 CHAPTER 7 LCMGUI REFERENCE MANUAL Configure Save Director PAES Here you configure the structure of the default directory where your LCModel results will be saved It will then be offered in the Control Parameters window The configuration is constructed by one or more successive elements that you select below Clicking on the green
166. ies in the column headed Extension are only the conventions used in this manual you are free to use others 9 6 EDDY CURRENT CORRECTION 115 9 6 Eddy Current Correction Eddy current correction ECC is automatically done by Philips if you set spectral correction to yes on the post proc page you can then skip Sec 9 6 This is recommended for single voxel not CSI data LCMgui users can skip Sec 9 6 and instead read Sec 6 3 3 ECC requires that a second set of raw time domain data be acquired with each set of in vivo data This second set called the Unsuppressed Data is acquired with exactly the same protocol but without water suppression It must be acquired from the same voxel immediately before or after the water suppressed acquisition Details are in 10 You can have LCModel do ECC as follows DOECC LOGICAL do eddy current correction TRUE to do the eddy current correction in LCModel Default DOECC F FILH20 CHARACTER 255 the name of the file containing the time domain Un suppressed Data e g pat_777_77 h2o0 This file has the same format as the RAW file 9 7 Prior Phasing Information LCMgui users with GE Probe data can skip Sec 9 7 the settings are done for you GE Probe users using the SAGE LCModel interface must make the entries specified in Sec 7 2 5 3 LCModel spends much of its time in an elaborate Preliminary Analysis to get good starting values for the phases and
167. ifted away from PPMSCA and PPMPHA makebasis auto in illustrates the case where you shift Cho main peak at 3 19 to approximate GPC or PCh main peaks at 3 22 using XTRASH 3 22 3 19 0 03 Another example is XTRASH 0 035 to shift NAA 2 01 ppm to ap proximate NAAG 2 045 ppm Default XTRASH 0 0 REAL DEGPPM approximately is only needed when AUTOPH T and only when the 1st order phase correction is large as with old Marconi Picker PRESS data i e when DEGPPM gt 180 PPMSCA PPMPHA This is 21 deg ppm when formate amp DSS markers are used It must then be a rough approximation e g to within 21 deg ppm of DEGPPM to resolve the ambiguity in large 1st order phase corrections You must have DEGPAP lt 200 but it should be much less shift the time domain data if necessary Default DEGPAP 0 0 which is fine if DEGPPM is less than about 15 Otherwise input DEGPAP equal to your approximate value of DEGPPM REAL 2 ppm range for offset correction This does not affect the LCModel results it is only for aesthetics If a Basis Spectrum has a con stant offset then the LCModel Baseline will compensate this with the same offset This can look strange if the Baseline has to move higher cutting through metabolite peaks or move way below the data spectrum it would also falsify the output S N Sec 2 4 MakeBasis will adjust the offset so that the average offset over the ppm 96 CHAPTER
168. igure and no decimal point in the number is not in this manual it is in the file HOME 1cmodel doc figures pdf You should output this on a printer with PostScript capability as was done with this manual which is in the same directory It is convenient to have these figures bound separately from this manual so that the two can be read side by side 1 Fig 2 means Fig 2 in Ref 1 1 4 Acknowledgments I thank the Biomedizinische NMR Group in Gottingen especially Wolfgang Hanicke for providing much data during the development of LCModel The first 10 000 spectra were processed with LCModel by them 2 I thank Jens Frahm for permission to use the test data in this manual I am very grateful to many other users of LCModel for helpful comments which have significantly contributed to the development of LCModel I thank the groups at GE Medical Systems Toshiba Medical Systems Marconi Bruker Medical Varian Philips and Siemens Medical Solutions for providing much information on their data structures I am especially grateful to Rolf Schulte Mu nich for guidance and GAMMA routines for simulating model metabolite spectra to Jim Murdoch Cleveland for detailed information on Picker and Philips data con version and to JBob Brown Fremont and Timo Schirmer Munich for essential information on GE data conversion I thank the GE Marconi and Bruker groups and Else Danielsen Copenhagen Thomas Michaelis Gottingen and Petra Pouw
169. ile that you selected in Step 1 in Sec 7 1 LCMgui does this by executing HOME 1cmodel bin2raw where bruker ge5 gelx marconi philips siemens toshiba varian or other depending on the type of data file that you specified in the menu in Fig 6 3 Actually LCMgui first attempts to execute HOME 1cmode1 my bin2raw If my bin2raw does not exist then it executes bin2raw This gives you the opportu nity of copying bin2raw to my bin2raw and modifying my bin2raw to your special purposes Your my bin2raw will be preserved since only bin2raw is overwritten during LCMgui updates Normally you would produce other my bin2raw and use the Other button in Fig 6 3 for your special purpose processing HOME 1cmodel other bin2raw pro vided with LCMgui does no data conversion it simply assumes that your selected file has already been converted to LCModel RAW format Sec 5 2 However the comments in this script also specify what you must do to prepare your particular type of data files After you make your special other my bin2raw then in your next LCMgui session select your non standard file with the File Selector Your previous LCMgui session probably did not use the file type Other and you will then get the error window in Fig 6 2 Click Retry and then select Other in the window in Fig 6 3 to use your customized script in other my bin2raw The scripts bruker bin2raw and philips bin2raw provide furth
170. in2raw to handle your file types the instructions in the File Selector will still incorrectly tell you that you must select an LCModel RAW file 7 4 5 Optional Preprocessor This is Step 3 in Sec 7 1 It gives you the opportunity of modifying the Control Parameters or the RAW file computed from your standard data conversion in Step 2 Only files in the directory HOME 1cmodel preprocessors will be available as Preprocessors in Figure 7 15 Here bruker ge5 gelx marconi philips siemens toshiba varian or other depending on which type of data you selected in Fig 6 3 Your own Preprocessor scripts must also be in this directory HOME 1cmode1 doc preprocessors README txt explains how to make your own Preprocessor You can check the Preprocessor currently being used by clicking Advanced Settings in Fig 7 2 The current Preprocessor is in parentheses next to Change Preprocessor In Fig 7 6 there is no Preprocessor is being used You can then change the Preprocessor as follows 74 7 4 5 1 CHAPTER 7 LCMGUI REFERENCE MANUAL Click on Change Preprocessor to produce the window in Fig 7 15 If there is no Preprocessor script in the directory then you will be told this and given the chance to continue The currently used script is shown in the green field under Selected File The original default is no Preprocessor step the green field would then be blank If you want to use a
171. ing a phantom that contains none of these metabolites see Sec 11 6 warning One or more of the NUSE1 Metabolite Names in the array CHUSE1 is not in the Basis Set The Preliminary Analysis will not use all the metabolites named in CHUSE1 and this could weaken the Preliminary Analysis Hints See Hints in MYBASI 7 Also make sure that your Metabolite Names agree with the Allowed Names in Table 8 1 see the discussion of METABO The field strengths in your in vivo data and your Basis Set differ by more than 20 FATAL or 5 20 warning LCModel will try to correct by bandwidth conversions but warning may mean poor modeling of coupled resonances Hints If HZPPPM is correct in your input and in your Namelist SEQPAR of your Basis Set then you need to get an appropriate Basis Set ERROR The echo times ECHOT in your in vivo data and your Basis Set differ by more than 3 ms Hints If ECHOT is correct in Namelist SEQPAR of your data and of your Basis Set then you should use an appropriate Basis Set ERROR Your in vivo data were acquired with STEAM and your Basis Set with PRESS or vice versa Hints If SEQ is correct in Namelist SEQPAR of your data and of your Basis Set then you need an appropriate Basis Set FATAL You have BASCAL T and at least one of the following requirements was not met NCALIB gt 1 CHCALI must be defined in your input FATAL You have BASCAL T and have input the same Metabolite Name for CHBCAL and one
172. ing for NAMEAC 9 9 ONE PAGE OUTPUT 121 9 9 3 2 Dimensions The overall page dimensions for your One Page Output are usually best set by inputting PGNORM as US or A4 Sec 5 3 6 Only if you leave PGNORM are the following two Control Parameters used e g for specifying non standard page dimensions PAGEHT REAL the height of your page in cm Default PAGEHT 27 9 i e 11 inches PAGEWD REAL the width of your page in cm Default PAGEWD 21 0 LANDSC LOGICAL Landscape TRUE if the ppm axis is determined by the page height Default LANDSC T this is almost always correct i e if you think of the ppm axis as horizontal then you can think of the page being rotated 90 and PAGEHT now defining the width XLEFT XRIGHT REAL the margins in cm on the left and right of the page i e the ppm axis will shortened by these amounts Defaults XLEFT 1 3 XRIGHT 1 3 YTOP YBOTT REAL the margins in cm at the top and bottom of the page Defaults YTOP 1 5 YBOTT 1 3 9 9 3 3 Lettering Lettering dimensions are specified in points about 1 72 inch PTTITL REAL the size of the title at the top of the page Default PTTITL 11 PTLABL REAL the size of the axis labels Default PTLABL 7 8 PTOUTP REAL the size of the lettering in the tables Reducing PTOUTP will increase the space for the ppm axis each line in the tables still has the same number of characters Inputting a sma
173. ion Philips users can skip Sec 6 3 3 you should have Philips do the eddy current correction ECC as specified in Sec 3 7 6 Users with Toshiba Marconi Picker data or GE Probe P files can skip Sec 6 3 3 since eddy current correction ECC is done automatically Other users only have to set this as default the first time for future analyses In the fourth row in Fig 6 4 a mouse click switches the button for ECC between active blue and inactive no ECC Blue usually indicates a button that is switched on ECC is recommended for single voxel data but not for CSI where it is usually ineffective You then must also have acquired an unsuppressed water reference spectrum from the same voxel immediately before or after acquiring your water suppressed spectrum With a CSI data set all of the voxels in the unsuppressed reference data set must be the same as those in the suppressed data set If you have this reference data file be sure to make the ECC button blue In rare cases particularly if the water suppression is weak and the unsuppressed water reference not very good you may see that ECC obviously amplifies the noise in the high ppm part of the spectrum If this amplified noise extends into the Analysis Window you should switch off ECC for this spectrum only 6 3 4 Do water scaling Water Scaling uses the unsuppressed water reference signal as used above for ECC to estimate absolute concentrations Section 10 2 2 discusses the signifi
174. is extreme shift really is necessary You can also see this from the Absolute Value Plot If this is so then input PPMCEN or SHIFMN 2 or SHIFMX 2 12 2 STANDARD LCMODEL DIAGNOSTICS 171 6 N 00 o 10 11 13 14 15 16 17 18 19 20 21 info The CCF with the second set of 6 functions specified by PPMREF 2 would cause an extreme shift of SHIFMN 2 Hints This is often caused by the spectrum climbing steeply towards the incompletely suppressed water peak or possibly a strong lipid peak A new CCF will be computed with NREFPK 2 reduced by one to eliminate the 6 function closest to water warning The action in STARTV 6 has been repeated until there are only 2 6 functions left usually NAA and Cr and the extreme shift is still being called for by the CCF Hints Either the true referencing shift is extreme reset PPMCEN SHIFMN or SHIFMX or your spectrum has very little structure or extreme residual water or lipid peaks more common with CSI If the referencing looks reasonable and not like PLOT 17 in HOME 1cmodel doc figures pdf then LCModel has managed anyway and there is no problem Otherwise Sec 11 3 can help warning The FWHM of a CCF extends to the right beyond the whole spectral window Hints As in STARTV 5 warning The FWHM of a CCF extends to the left beyond the whole spec tral window Hints As in STARTV 5 or the residual water peak is very strong ERROR You input
175. is not accepted during installation This suppresses printing and display but not the archiving of the plots in the PS files This can be useful if you are processing many spectra e g a multi voxel data set and want to view the results later from the archive You can later collect all of the PS files into one file for convenient viewing using Sec 11 4 5 7 5 Fine Points 7 5 1 gui defaults Resource File LCMgui stores all of your default settings for future sessions in HOME 1cmode1 profiles gui defaults where 7 is your currently se lected User Profile 1 by default Each line consists of a one word index e g i_displayCommand whitespace a value e g gv orientation seascape and nothing else on the line In exceptional cases listed below you may wish to edit this file You should only edit the value and only as specified below 7 5 1 1 Starting Over If you hopelessly corrupt gui defaults you can always recover by deleting gui defaults When you start LCMgui the next time it will be as with a new installation You first click Repeat Test Run then install the license unnecessary with Linux and then proceed with data analysis All of your previous defaults and mistakes will be forgotten 7 5 1 2 Protecting your Settings You can make the initial LCMgui runs and optimize the default settings for Normal Users During any future LCMgui session the user can modify any of your settings Normally LCMgui wil
176. is produces the test auto basis file and a PostScript file basis auto ps both in the directory HOME 1cmodel test output Output basis auto ps on a PostScript printer There is one plot for each metabolite with this test data only three The top line of each plot contains the metabolite name the phase corrections if AUTOPH T and the Scale Factor if AUTOSC T MakeBasis multiplies TRAMP by this Scale Factor in test auto basis so that all of your Basis Spectra are automat ically scaled consistently with each other All Basis Spectra are scaled to match the formate standard of the first metabolite in the IN file Cho in this example thus its Scale Factor is identically 1 8 6 2 Judging the Basis Spectra You should always carefully check the plots in the PS file output by MakeBasis The phasing and referencing should look reasonable and the quality should be com parable to that of the test runs in basis ps and basis auto ps in the directory HOME 1cmode1 test output If you use Auto Scaling or Auto Phasing you should run MakeBasis 3 times with the default PPMST PPMEND of 4 2 0 2 but also with 8 54 8 34 and 0 1 0 1 to zoom in on the formate and DSS peaks The formate and DSS peaks should be at least as well defined and phased as in the test run basis auto ps where only 100 mM formate and 1 mM DSS were used 8 6 3 RAW File These files have exactly the same format specified in Sec 5 2 for the RAW files
177. is to make your input PPMSHF possible Hints Adjust PPMST PPMEND PPMSHF or your bandwidth TWOREG two regularizors Manages the Final Analysis finding the solution with the best pair of regularization parameters ag and ag in 1 Eq 4 1 2 FATAL An essential unregularized nonlinear least squares analysis has failed Hints As in RFALSI 3 FATAL RDALPB lt 1 0 or ALPBST lt ALPBMN or ALPBST gt ALPBMX Hints All of these are Hidden Control Parameters do not input them warning A less than normal extent of lineshape smoothing has been ex plored The regularizor might be suboptimal Hints Check for very poor data or spectral resolution FATAL A regularized solution within the acceptable range of the confidence region in 11 could not be found Hints Check for very poor or erroneous data e g that would require a wild Baseline Do not modify Hidden Control Parameters such as PRMNMX FATAL Both PRMNMX 2 1 and PRMNMX 2 3 exceed PRMNMX 1 2 Hints Do not input these Hidden Control Parameters warning An important regularized analysis has failed to properly converge The final results may be affected 12 2 STANDARD LCMODEL DIAGNOSTICS 173 10 11 12 13 14 15 16 17 Hints Check for very poor data or spectral resolution info A regularized analysis in a refinement step has failed to properly converge The final results should not be seriously affected Hints Working Hard
178. ite concentrations in a series of mea surements Since there is no user interaction the results are user independent thus improving objectivity and exchangeability of the results within and between labora tories 1 3 Conventions amp Notation Metabolite abbreviations are used throughout and are defined in Table 8 1 As already used above 1 means reference 1 in the Bibliography at the end of this manual Special terms such as Normal User are italicized the first time that they appear and capitalized thereafter Less obvious special terms are in the Index at the end of this manual and are therefore not cross referenced when they appear in the text 1 4 ACKNOWLEDGMENTS 13 File names and LCModel input output are written in the teletype style as figures pdf Your home directory is specified by HOME although you can usually abbreviate this with a tilde Unix means Sun SGI or Compaq DEC in the few cases where they behave differently from the Linux version Concentrations should be labeled mmol per Kg wet weight We use the shorter in correct abbreviation mM The actual mM is the mmol per Kg wet weight multiplied by the specific gravity of the tissue typically 1 04 in brain We use the term directory and not folder Further conventions for more advanced users are defined in Sec 5 1 1 3 1 Figures Figure 6 2 or Fig 6 2 refers to Fig 6 2 in Chap 6 of this manual PLOT 6 PLOT rather than F
179. l Parameters You can change the Control Defaults file currently being used as follows e Click on the Advanced Settings button in Fig 7 2 to produce the menu in Fig 7 6 e Click on Change Control Defaults File to produce the window in Fig 7 10 If there is no Control Defaults file in your directory then you will be told this and given the chance to continue e The currently used file is shown in the green field under Selected File 68 CHAPTER 7 LCMGUI REFERENCE MANUAL New User Profile Nane A In the green field below enter the new User Profile name where your default settings are to be saved In the window at the bottom are your User Profile names that already exist Ifyou enter one ofthese names in the green field then you will overwrite this old User Profile with your current one New tename ERE Click on Save below to create the User Profile that you entered in the green field above Click on Cancel to continue using your old User Profile without creating a new one In either case your default settings will be saved to the chosen User Profile before exiting normally Save Cancel 3 bruker chla test test ds ttsh Figure 7 11 At the end of a session you can select a new User Profile where all of your settings from the session will be saved The original default is no Control Defaults File the green field would then be blank e As explained in Fig 7 10 you ca
180. l differing from mM by typically 20 or more much more with TE gt 35 ms or TR lt 3000 ms However if relaxation times and acquisition parameters do not change much during a study then this factor may not change much either and you may still get reasonably consistent measures of absolute con 10 2 WATER SCALING 129 centrations in your institutional units that are useful at detecting e g increases or decreases in pathologies WCONC ATTH20 ATTMET REAL the NMR visible water concentration mM in the voxel This is 55556BuRr in 28 where typical values were 35880 for white matter and a 21 larger 43300 for gray matter Default WCONC 35880 0 There is no reason to measure phantoms but if you did you would obviously use WCONC 55556 REAL the attenuation of the NMR visible water signal due to additional relaxation effects and other effects such as transmitter amplifier leakage 22 p 337 or the extra attenuation in the Gain of the unsuppressed water spectrum In this last case you must include the factor GaiNnunsuppressed G Ol suppressed in ATTH20 The default value of ATTH20 is 0 7 so for example if the above gain ratio were 0 5 you would input ATTH20 0 35 Thus ATTH20 must contain all corrections that affect the unsuppressed water reference but not the water suppressed signal Equal effects in both unsuppressed and suppressed signals cancel out because the un suppressed signal is in t
181. l spectrum To use this you must read Sec 9 3 SPTYPE breast 8 for spectra with only lipid and water and possibly choline signals PLOT 10 shows a breast spectrum but this is not only for breast To use this you must read Sec 9 3 SPTYPE only cho 2 for spectra where you only want to estimate choline e g with breast where fat suppression has distorted the lipid signals around 1 3 ppm To use this you must read Sec 9 3 SPTYPE liver 11 for spectra with only lipid and water choline signals and glycogen PLOT 11 shows a liver spectrum To use this you must read Sec 9 3 SPTYPE mega press 3 for estimating GABA using MEGA PRESS To use this you must read Sec 9 4 PLOT 12 shows an example SPTYPE cs when you expect mainly cerebrospinal nul witha ree csf when y pect mainly brospinal fluid CSF with all th 102 9 2 MUSCLE SPECTRA 103 landmarks NAA cholines and creatines weaker than Lac amp Gle PLOT 13 SPTYPE version5 is for consistency with very old versions Sec 11 17 SPTYPE tumor when the NAA singlet may be weak compared to choline or cre atine or the lipid or macromolecule signals around 1 3 or 0 9 ppm So this is not only for tumors PLOTs 6 amp 7 The following except 2 are automatically set 1 Since there is often only one useful landmark at 3 22 ppm SHIFMN 2 0 07 and SHIFMX 2 0 07 to prevent Cr from being shifted to 3 22 If your
182. l store these new settings in gui defaults for use as defaults for the next session However you can prevent LCMgui from storing changes to critical settings by changing the value of i_preserveConfigurationin gui defaults 76 CHAPTER 7 LCMGUI REFERENCE MANUAL from 0 to 1 i_preserveConfiguration 1 Some of the critical defaults that are protected by the above setting are e The structure of the Archive Directory for LCModel results and which types of output files will be saved there e The default BASIS file e The print and display commands e The pathnames of the files with the default Control Parameters the Pre processor and the Execution Script The license data In this way users have full freedom to change everything for their present session but they and others are protected from the most critical of their changes in future sessions 7 5 1 3 Data Filename Filters The File Selector window in Fig 6 1 will only display filenames that match the fil ter which is specified by the values i_fileType where Bruker Ge5 GeLx Marconi Other Philips Siemens Toshiba or Varian You can temporarily enter a new filter in the entry below Files in Fig 6 1 However you can perma nently change this in gui defaults For example you can specialize ifileTypeSiemens to i_fileTypeSiemens rda 7 5 1 4 Window amp Font Size You may find the LCMgui windows too big The best way to reduce them is to in cre
183. lained below Default RGBRAT 0 999 0 0 0 0 i e red The three elements of RGBBOL amp RGBRAT specify the fractions of red green amp blue respectively For example 1 0 0 0 0 0 is red 0 0 0 0 1 0 is blue 0 0 0 0 0 0 is black not 1 1 1 These fractions must always be in the range 0 0 1 0 I input 0 999 instead of 1 0 since Namelist input can have slight rounding errors 9 9 2 Diagnostics Table Diagnostics of type warning are output in boldface in the color specified by RGBBOL RGBERR REAL 3 RGB for errors RGB specification for the color of diagnostics of severity ERROR and higher Default RGBERR 0 999 0 0 0 0 i e red 9 9 3 Plot The Control Variables in this subsection allow you to tailor the plots e g for slides or publications 9 9 3 1 Individual Metabolite Contributions You can get separate plots for the contributions of individual metabolites to the data as in PLOTs 3 amp 4 in HOME 1cmodel doc figures pdf This can be useful for a detailed evaluation e g of the compatibility of your data with the Basis Spectra NEACH INTEGER the number of metabolites for which individual plots are to be made Default NEACH 0 NAMEAC CHARACTER 60 6 the first NEACH elements contain the Metabolite Names for the individual plots If NEACH gt the total number of Basis Spectra in the analysis then individual plots are made for all metabolites and you do not need to input anyth
184. larized lineshape function which accounts for field inho mogeneity etc However the lipid or MM peaks can have highly variable deviant lineshapes that have nothing to do with the normal metabolite lineshapes These stronger lipids amp MMs would dominate the determination of the lineshape func tion The normal metabolites would then have this incorrect lineshape causing significant errors in their concentration estimates Therefore the lineshape function is not used for the simulated lipids and MMs Their lineshape is adjusted as a crude Gaussian Lorentzian model The two major signals near 1 3 amp 0 9 ppm are given more flexibility six parameters by the two Lipl3a amp Lip13b and by MMO9 amp Lip09 respectively Concentration Ratio Priors Even with the above two steps the analysis is still seriously over parameterized It is also necessary to impose soft nonlinear constraints on the ratios of lipid components to Lip13a Lip13b and of the MM components to MMO09 They stabilize the analysis considerably and they help to separate Lip09 from MM09 and Lip20 from MM20 by the chains of constraints from the more accurate Lip13a Lip13b to Lip09 etc and from MM09 to MM20 etc They are specified in Sec 11 8 Flexible Baseline The Baseline has been given the necessary extra flexibility to correct for the inevitable inaccuracies particularly in the crude lipid amp MM lineshape models 11 8 Concentration Ratio Priors You c
185. ld have PPMEND lt 1 0 to give sufficient baseline to the right of the lipid signals around 1 3 amp 0 9 ppm warning With SPTYPE lipid 8 breast 8 or liver 11 and PPMST gt 5 0 you should have PPMST gt 8 0 to give sufficient baseline to the left of the water signal warning With SPTYPE liver 11 and PPMST lt 5 0 you should have PPMST 4 0 to give sufficient baseline to the left of Glycg but to avoid the tail of the residual water signal warning With SPTYPE breast 8 and PPMST lt 5 0 you should have PPMST between 3 8 and 4 0 to give sufficient baseline to the left of Cho but to avoid coming too close to the tail of the residual water signal warning With SPTYPE lipid 8 and PPMST lt 5 0 you should have PPMST between 3 4 and 4 0 to give sufficient baseline to the left of the lipid signals but to avoid coming too close to the tail of the residual water signal warning With SPTYPE only cho 2 you should have PPMST between 12 2 STANDARD LCMODEL DIAGNOSTICS 165 3 8 and 4 0 and PPMEND between 2 6 and 2 8 to give sufficient baseline to the left and right of Cho but to avoid coming too close to the tail of the residual water signal or the lipid signals 23 FATAL With SPTYPE lipid 8 breast 8 or liver 11 the Hidden Control Parameter IPOWRG must be 1 or 2 Hints Do not input IPOWRG 24 FATAL You must have the Hidden Control Parameter NBCKMN
186. lename ps dir filename_s12_3 4 ps dir filename neither of the above dir filename_s12_3 4 So the identifier is of the general form CHSLIC CHROW CHCOL where the are slice number row number amp column number respectively You specify the identifier as follows CHSLIC CHARACTER 9 Default CHSLIC s1 CHROW CHARACTER 9 Default CHROW _ underscore CHCOL CHARACTER 9 Default CHCOL minus sign 11 4 3 2 Inconsistencies between LCModel and Scanner Labeling amp Dis plays LCModel strictly follows the DICOM standard The order of voxels in a slice is from left to right top to bottom Thus the upper left voxel labeled 1 1 is followed by 1 2 and the rest of row 1 followed by the left voxel of row 2 labeled 2 1 and so on With Siemens data the LCModel voxels will be those of the CSI slice They will also be those of the Siemens reference image provided that the CSI slice has not been rotated with respect to the reference image Sometimes scanner displays are transposed i e the rows and columns are interchanged Sometimes the labeling is also changed from row column to column row In some early scanner versions the number of rows and columns is interchanged in violation of DICOM and it is necessary to also interchange the total number of rows and columns in Fig 6 4 or if LCMgui is not used Sec 5 3 2 before starting LCModel 11 4 MULTI VOXEL DATA SETS 139 11 4 4 Output fo
187. les iaaiGeBS065 ge Only for Multi Voxel or Multi Channel data files Total data Columns m Selecting Columns m through ig for Preview and then Analysis Total data Rows m Selecting Rows E through ur for Preview and then Analysis Total data Slices MI Selecting Slice EN for Preview and then Analysis Advanced Settings _Run LOModel Preview Data Reload Data Exit LCMgui Figure 7 2 The Control Parameters Window 7 3 2 BASIS file Your BASIS files should go into the directory HOME 1cmodel basis sets Then LCMgui normally automatically selects the correct Basis Set based on the information in the scanner data files In some cases information is missing e g TE in older GE spectrum G files and you should check the entry next to BASIS file in the Control Parameters window Fig 7 2 and click the Change BASIS button if necessary If you are using only BASIS files supplied by me then you can skip the rest of Sec 7 3 2 LCMgui attempts to automatically select the correct BASIS file from the default Basis directory the Basis directory from the last session It first attempts to extract TE and the localization sequence STEAM PRESS from your data headers It then attempts to find the corresponding BASIS file whose filename must satisfy the following rules TE The filename must contain the string _teNNN_ including underscores at each end where NNN is the integer value of TE in ms An example is gamma_press_
188. ll value like PTOUTP 0 001 will make the tables invisible e g for a slide or figure Default PTOUTP 7 8 The vertical space from one line of lettering to the next is given by the following factors multiplying the lettering size e g a factor of 1 0 is a little closer than single spacing a factor of 2 0 is almost double spacing RHTITL REAL RHLABL REAL axis label spacing Default RHLABL 1 2 RHOUTP REAL table spacing Default RHOUTP 1 3 RHVERS REAL title spacing Default RHTITL 1 5 REAL spacing of the lines with your laboratory name and the version number of LCModel at the top of the page Default RHVERS 2 9 9 3 4 Curves The following dimensions are all in cm 122 WDLINE RGBLIN DSHPAT SUBBAS 9 9 3 5 XSTEP NSUBTK CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION REAL 6 these six elements give the curve thicknesses for the 1 fit to the data 2 Baseline the plot of the Baseline can be suppressed e g for figures with WDLINE 2 0 3 data 4 Residuals 5 axes and frame 6 fine grid lines The vertical grid lines will be suppressed if WDLINE 6 0 e g for a slide or figure Defaults WDLINE 06 01 005 01 04 005 REAL 3 6 RGBLIN J K J 1 2 3 is the RGB color specification for the curves of Type K using the Type numbering of WDLINE Default RGBLIN 1 1 0 999 all other elements are 0 0 Thus line Type 1 the fit to the data is red
189. lt 15 are in boldface blue e Of course averages over a group of LCModel analyses of similar spectra can significantly reduce the uncertainties Section 11 1 1 specifies how to compute these average concentrations 15 16 CHAPTER 2 ONE PAGE OUTPUT The abbreviations of the metabolites are in the 4th column Table 8 1 defines most of these Metabolite Names The corresponding absolute concentrations are in the 1st column Usually the units of the absolute concentrations are unknown and only concentration ratios are meaningful These ratios are given in the 3rd column whose heading Cr indicates that the ratio is relative to creatine plus phosphocreatine Only when the Basis Spectra and your input data are consistently scaled with each other are the absolute concentrations meaningful for example you can use Water Scaling Sec 6 3 4 amp Chap 10 2 1 1 Upper Part The Concentration Table usually is split into two parts by a horizontal line The upper part contains the usual low molecular weight metabolites of interest NAAG is quite difficult to resolve from NAA and PCh is very difficult to resolve from GPC as is PCr from Cr With lower quality spectra or long TE Glu and Gln are also often difficult to resolve The last three rows of the upper part of the table are the sums of these concentration pairs The sum GPC PCh is much more accurate than the individual concentrations as can be clearly seen from the much lower SD f
190. lt to resolve from Lip13 Lip16 lipid signal around 1 6 ppm At most resolutions this is considered part of Lip13 Thus for fat your would probably use Lip16 Lip13 or L16 L09 L13 in the output table Lip21 lipid signal around 2 02 ppm Lip23 lipid signal around 2 23 ppm this often cannot be resolved from Lip21 Lip28 lipid signal around 2 75 ppm Lip41 lipid signal around 4 1 ppm Lip43 lipid signal around 4 3 ppm Lip41 amp Lip43 are weak and only seen with high resolution and signal noise Water around 4 65 ppm if PPMST 8 0 This value is only meaningful when the spectrum is unsuppressed i e with the One Spectrum Method Lip53 lipid signal around 5 3 ppm Lip52 lipid signal around 5 2 ppm This can almost never be resolved from Lip53 and only the sum Lip53 Lip52 should be used Cho signal from choline containing compounds around 3 2 ppm omitted when SPTYPE lipid 8 PLOT 10 in HOME 1cmode1 doc figures pdf is from breast Glycg formerly called Glyu only with SPTYPE liver 11 signals in the region 3 5 3 9 ppm Part of these may be due to glycogen Clearly the sums like Lip16 Lip13 are used in the output because they are more accurate than the parts Lip16 and Lip13 In the Concentration Table the meanings of Conc 1st column and the resonance area ratios 3rd column depend on your input of PPMST and DOWS T Water Scaling With Water Scaling the Conc column has
191. ly PPMGAP 1 1 5 1 and PPMGAP 2 1 4 2 Default PPMGAP 20 1 E37 i e no gaps 148 CHAPTER 11 FINE POINTS For better labeling of the wide ppm axis Sec 9 9 3 5 you might input NSUBTK 5 XSTEP 0 5 This could be useful for Glc quantitation at 4 7T and above 16 Fig 9 At lower fields water suppression would probably also suppress the Glc signal at 5 23 ppm It could also be useful for metabonomics of body fluids at high resolution where there are many signals above 5 ppm Quantitation of phenylalanine in brain at lower fields is probably hopeless because of its extremely weak signal and overlap around 7 4 ppm with homocarnosine and a relatively strong poorly defined macromolecule signal Also NAA has a signal around 7 8 ppm at 37 C 11 12 Multimodal Lineshapes The finite convolution in the last sum in 1 Eq 1 is a very flexible model free correction for lineshape The convolution should extend over a wide enough range to cover the linewidth but not so far that it wastes computer time and reduces numerical stability This extent is determined by the FWHM of the lineshape estimated in the Preliminary Analysis as follows RFWHM REAL the convolution extends over a range of RFWHM FWHM 2 in both directions Default RFWHM 1 8 The FWHM estimate from the crude Preliminary Analysis can be too low if the real lineshape has several peaks as with the extreme eddy current effects in 1 Fig 3b This is obvious fro
192. m done These reference data are automatically included with the water suppressed data in a single voxel GE Probe raw P file or Marconi Picker file and the two required LCModel RAW files see Sec 9 6 can be made In other cases you have to organize this yourself You put the unsuppressed reference data in a separate file FILH20 You then add DOECC 4 FILH20 as specified in Sec 9 6 to your CONTROL file In addition to improving the Baseline and lineshapes ECC usually greatly reduces the zero order phase correction Section 9 7 1 shows how you can experimentally investigate whether this correction is small If it is then this prior knowledge can speed up and improve the LCModel analysis especially of poor quality spectra With single voxel data add the following Control Parameter to your CONTROL file to im pose this prior information SDDEGZ 6 The Index at the end of this manual tells you where to find the discussion of other Control Parameters such as SDDEGZ With CSI data ECC may be less effective and the phases highly variable Do not input SDDEGZ let the Bayesian learning in LCModel set SDDEGZ For Bruker users only In ParaVision choose the option to have ECC done Do not do ECC again in LCModel Without LCMgui you first correct these for Bruker s digital filter using Bruker s convdta Do not input SDDEGZ since the zero order phase correction is often not small In rare cases particularly if the water suppressi
193. m the poor fit to the side peaks and the non random residuals You should discard such a spectrum However you can nevertheless rerun with an increased RFWHM say to 3 0 or more if necessary 11 13 Relaxation amp Shift Priors The last sum in 1 Eq 4 imposes normal prior probability distributions for the broadening parameters ye and the shifting parameters ez in 1 Eq 3 These priors are mainly to stabilize the nonlinear least squares analysis when it is far from the optimum or when a metabolite has nearly zero concentration Normally you should not change these Control Parameters Only if your metabolites have very unusual differences between their in vivo and in vitro spectra would it be worthwhile to adjust these Control Parameters The priors for each metabolite in 1 Eq 4 are defined by three quantities The last four characters of the Control Parameter names tell which of these three quantities are being modified SDSH SD of shift These are for the ey They specify the uncertainty in the refer encing in ppm between in vitro and in vivo 11 14 DETAILED OUTPUT 149 EXT2 expectation of 1 T2 Control Parameters with this ending modify yp in 1 Eq 4 These Control Parameters specify the increase in 1 T2 in s in going from the in vitro Basis Spectrum to in vivo SDT2 SD of 1 T2 These are for the o y the units are also s The SDT2 amp EXT2 Control Parameters are scaled by the factor
194. makes LCMgui very flexible At the same time LCMgui remembers most settings from the previous run so if you make the first analyses after installation Normal Users with limited knowledge can often analyze their data with your configuration You can even pre vent Normal Users from permanently changing your most important settings using Sec 7 5 1 2 7 1 1 Limitations Unlike LCModel which was written by one person and is quite self sufficient LCMgui combines free software from several sources and depends on your particular window manager and system With Linux some window managers especially old ones may on rare occasions require that you manually move and resize some windows Some times you get a tiny window smaller than an icon which you have to find usually toward the upper left click on and then resize it is best to then exit and start LCMgui again Most displays have sufficient resolution for LCMgui The display must have at least 1024 x 768 pixels better 1600 x 1200 or 1280 x 1024 800 x 600 pixels are not enough If the LCMgui windows are only slightly too large then you can try reducing the font size Sec 7 5 1 4 7 2 Installing LCMgui As specified in Chap 4 downloading from the WWW site on Page 2 and installing and uninstalling are very simple Do not work as root for example this could set file access privileges so that LCModel updates by the user do not work properly The entire package goes into HOME 1c
195. mand for displaying or printing PostScript files gv or evince from your Linux distributor s repository gv written by one person more than 20 years ago is better than evince for PostScript files evince is better for PDF files It is easy to install gv from the Ubuntu repository With evince you can put an icon on the toolbar for rotating the plot left Some Linux distributions e g Ubuntu have missing fonts or files that cause warn ings with evince and the others Usually these do not affect the output and can be ignored However LCMgui will ask for a corrected display command If the display or printing was successful you can click Use Old Command You can later suppress these warnings by using Sec 7 4 6 to change the display command e g to evince 2 gt dev null The test runs will automatically output the One Page Output four pages in this case These should approximately agree with PLOTs 1 4 in HOME 1cmodel doc figures pdf 4 2 2 PostScript Printer You can instead output to a printer Most printers can print PostScript files If your default printer can you would probably enter lpr in the green entry field in Fig 4 1 For another printer e g called mypr you would enter a command like lpr Pmypr 30 CHAPTER 4 INSTALLATION AND TEST RUNS 4 2 3 No Printer or Display If neither printing nor display is possible then enter the harmless command touch in the green field in Fig 4 1 LCMgui outputs
196. mation is very accurate and LCMgui uses this However older and newer PRESS sequences have very different values For backwards compatibility the older values are used by default Try a few analyses with and without the Preprocessor new press The wrong choice should yield very poor phasing If in doubt send me two or three dump files with high quality spectra 7 3 BASIC SETTINGS AND USAGE 59 Philips Be sure to use Sec 3 7 6 Siemens Since a wide variety of Siemens files are accepted you may select totally irrelevant files e g with 2 bytes and they may be accepted This can be checked with Preview Data Sec 7 4 1 but it is also usually apparent from errors in the TITLE line in Fig 7 2 You can then click Reload Data in Fig 7 2 Further important settings are in the following sections 60 CHAPTER 7 LCMGUI REFERENCE MANUAL ll Control Paraneters dia You can change any parameters below When you are satisfied click on Run LCModel at the bottom Non Expert users with single voxel data can often click on Run LCModel immediately Tm am ASSE AASS O TAT presse TERASS SA TORRE SSA a SEL POZSGO7 Chierens Hostal Las Angeles Analyzing spectrum from 40 ppm down to 0 2 ppm BASIS file SP sp lcmodel basis sets ge_press_te35_95e basis _Change BASIS Do eddy current correction Do water scaling Save File types to directory Romelsplemadel savea Chigrens_Hospial Los_Ange
197. meters 20 acquisition time 38 ALEXT2 149 ALSDSH 149 ALSDT2 149 Analysis Window 17 21 39 LCMgui 47 Archive Directory LCMgui 61 archive LCMgui output 60 ATTH20 129 ATTMET 129 Auto Phasing 90 Auto Scaling 91 AUTOPH 93 AUTOSC 93 averaging concentrations 133 averaging data 145 BADELT 98 bandwidth GE at 3T 74 BASCAL 99 Baseline 22 Baseline offsets 95 BASIS file 98 naming rules LCMgui 60 BASIS Namelist 98 Basis Set 33 78 Basis Spectra 78 excluding 117 installing in LCMgui 55 table 149 BASIS1 Namelist 98 batch processing LCMgui 71 Bayesian learning 20 Big3 144 bin2raw LCMgui 72 breast spectra 102 BRUKER 36 Bruker data 24 36 40 58 62 130 calibrating Basis Spectra 98 calibration scanner 127 carbon 13C 152 CCF referencing 136 cerebrospinal fluid 102 CHBCAL 99 CHCALI 100 CHCOL 138 CHCOMB 118 chemical shift imaging 20 137 CHEXT2 149 CHKEEP 118 CHNOT2 142 choline spectra 110 CHOMIT 117 CHRATO 143 CHROW 138 CHSDSH 149 CHSDT2 149 CHSIMU 141 CHSLIC 138 CHUSE1 140 coefficient of variation 133 coherent averaging 145 color RGB specification 120 colors LCMgui 77 One Page Output 119 CONC 94 CONC3F 152 concatenating multi voxel PS files 139 Concentration Ratio Priors 143 concentration ratios 119 Concentration Sums 118 Concentration Table 15 179 180 CONCSC 94 CONREL 119 constrained regularization
198. mma separated value file by spreadsheet programs LCSV INTEGER LCSV 0 the default will suppress creation of this file LCSV 11 will make this file 5 3 7 3 PS PostScript File The PS file is output automatically It contains all the information in the One Page Output including the plotted curves You could archive this in the same directory as the TABLE file You can then view it with a PostScript viewer Sec 4 2 1 You can also archive it on paper using a duplex PostScript printer 5 3 CONTROL FILE 43 5 3 7 4 COORD Coordinate File If you want to make your own plots the COORD file contains the coordinates of all curves on the One Page Output in an easy to read and self explanatory text file FILCOO CHARACTER 255 coordinate file the name of the COORD file to be output e g home user1 1lcmodel saved pat_777_77 coord LCOORD INTEGER LCOORD 0 the default will suppress creation of this file LCOORD 9 will make this file 5 3 8 Phantoms Tests with in vitro data from model solutions are unrealistic and not always useful The accuracy and spectral resolution in the Basis Set must be higher than that in the data and this is often not the case here Broadening the in vitro test data also smooths them and this falsifies the statistical tests 11 in LCModel Even without smoothing or broadening the systematic errors in such data often exceed the normal scatter due to the noise and this
199. mmended a little freedom is better SDDEGZ lt 22 will increase the speed of the Preliminary Analysis by a factor of 4 This is the most common strength for imposing prior knowl edge SDDEGZ lt 45 will cause LCModel to start looking for o at DEGZER SDDEGZ gt 45 will not affect the starting points of the search in pa rameter space Default SDDEGZ 999 i e effectively no prior information REAL your expected value for the first order phase correction using the definition of DEGPPM in Sec 8 5 2 Default DEGPPM 0 REAL the standard deviation in your expectation value of DEGPPM in de grees ppm See the discussion of SDDEGZ above The 4 levels of strength with which you can impose this prior information in decreasing order are SDDEGP lt 0 4 fixes to DEGPPM with a standard deviation of 0 0 not 0 4 This is not recommended a little freedom is better SDDEGP lt 7 will increase the speed in the Preliminary Analysis by a factor of 3 This is the most common strength for imposing prior knowl edge SDDEGP lt 10 will cause LCModel to start looking for at DEGPPM SDDEGP gt 10 will not affect the starting points of the search in pa rameter space Default SDDEGP 20 i e only a weak bias toward reasonably small 1 values Do not use SDDEGP gt 20 it can destabilize the analysis DGPPMN DGPPMX REAL These define the lower and upper limits in degrees ppm of the range over which L
200. mobile lipids Magn Reson Med 46 615 618 2001 R Freeman A Handbook of Nuclear Magnetic Resonance Longman Essex 1988 M Mescher H Merkle J Kirsch M Garwood R Gruetter Simultaneous in vivo spectral editing and water suppression NMR Biomed 11 266 272 1998 R L O Gorman L Michels R A Edden J B Murdoch E Martin In Vivo detection of GABA and Glutamate with MEGA PRESS Repro ducibility and Gender Effects J Magn Reson Imaging 33 1262 1267 2011 175 176 9 10 11 12 13 14 15 18 BIBLIOGRAPHY M A McLean F G Woermann G J Barker J S Duncan Quantitative analysis of short echo time H MRSI of cerebral gray and white matter Magn Reson Med 44 401 411 2000 U Klose In Vivo proton spectroscopy in presence of eddy currents Magn Reson Med 14 26 30 1990 S W Provencher A constrained regularization method for inverting data represented by linear algebraic or integral equations Comput Phys Commun 27 213 227 1982 Available as cpc1 pdf on http lcmodel ca contin shtml B J Soher K Young A Bernstein Z Aygula A A Maudsley GAVA Spectral simulation for in vivo MRS applications J Magn Reson 185 291 299 2007 http scion duhs duke edu vespa I Tkac J Kim K Ugurbil R Gruetter Quantification of H NMR spectra of human brain using LCModel and spin simulation at 7 Tesla Proc Intl Soc Magn Reson Med 9 214 20
201. model of the user s directory However you can avoid continually entering the dot and everywhere use 1cmodel instead of 1cmodel by entering cd ln s 1cmodel 1cmodel 7 2 1 Multi User Installations One license allows an unlimited number of users to have LCMgui and LCModel installations An unlimited number of users can simultaneously run LCMgui and LCModel on the licensed computer You could install into a new account and have all users use this installation This would make installing upgrades simpler However it is better if each user has their own installation as with a browser so that their settings and archive of results are not overwritten by others This also allows them to decided if or when to upgrade Upgrades are free 7 2 INSTALLING LCMGUI 55 7 2 2 Starting LCMgui You start LCMgui with the command HOME 1cmode1 1cmgui If you set your PATH environment variable to include HOME 1cmodel then you should be able to start LCMgui simply with 1cmgui 7 2 3 First Test Runs LCMgui uses most of the settings from the previous run as defaults for the current run So your first runs are most important since the Normal User can then later successfully use LCMgui often simply by selecting the data file and clicking Run LCMode The most important initial settings at this stage are Output You set the print or display command in the initial test run Sec 4 2 2 and all future results will be automaticall
202. model new mv HOME 1lcmodel old HOME 1cmodel You can also use symbolic links Do not delete your old HOME 1cmodel instead install your update on top of your old installation You install an update simply by downloading from http l1cmodel ca 1cm test shtml and installing with the two commands tar xf lcm tar install lcmodel where is linux sun sgi or dec 58 CHAPTER 7 LCMGUI REFERENCE MANUAL The test run will automatically be done but you can then continue as before since only files that have changed in the update will be modified Your license and other configuration settings will not be changed Any files that you have added or archived in the HOME 1cmodel directory tree will also be preserved The only exception is that you should not manually modify any of the files originally delivered with LCMgui since these are overwritten by an update However this is normally never done and is never recommended in this manual be sure to exit normally from LCMgui with the Exit amp Save button in Fig 6 6 so that all of your update information is saved 7 3 Basic Settings and Usage The first runs are the most important since LCMgui uses your previous settings as defaults for the future This section shows the most important settings which can often allow you and inexperienced Normal Users to make future analyses with only a few mouse clicks For the first run you should follow Chap 6 including all t
203. n 11 to within the desired tolerances could not be found The regularizor could be inappropriate Hints Working Hard info ag has reached its maximum allowed value Hints Common with lipid spectra SAVBES saves information on the best solution found so far 1 ILLOGICAL Meaningless argument in call to SAVBES SETUP initially sets up regularizors priors and starting estimates for the nonlinear least squares analyses in 1 Eq 4 1 2 du 14 15 FATAL One of the Hidden Control Parameters DKNTMN RBACKG or FWHMST has been set nonpositive Hints Do not input these ERROR Ng would exceed its maximum allowable value 80 It is being reduced to this value Hints This should cause no problems but it could only occur with unusual values of the Control Parameters in SETUP 1 or a very wide Analysis Window Not used FATAL an eigenvalue decomposition to get the regularizor Rp failed Hints Never Happened FATAL The number of parameters exceeds the available storage Actually this is an ILLOGICAL since these limits should have already been checked Hints Never Happened FATAL Same as SETUP 5 FATAL FWHMBA lt 0 and it must be positive FATAL Same as SETUP 5 Not used FATAL A starting value is ag lt 0 or ag lt 0 Hints This can only occur if you have modified Hidden Control Parameters such as ALPBST ALPSST or DKNTMN FATAL Same as SETUP 5 FATAL Same as SETUP 5 FATAL The Hid
204. n da ELA ete Ae 83 2 Consistency ing oa 2 e eh as Lg Soe a aes 8 3 3 Acquisition Time and Bandwidth 8 3 4 Repetition Time e Eddy Current Correction 8 4 1 ECC with LOCMgui 2 00 8 4 2 ECC without LCMgui Plotting RAW Files with PlotRaw 8 5 1 Test Run of PlotRaW o e e 8 5 2 PlotRaw IN File o o e 8 5 3 PlotRaw Diagnostics 2 eee ees Running MakeBasis 2 8 6020 236 Sob paa a ad 8 64 AutosSc aline cas ia a 4 ak A aa 8 6 2 Judging the Basis Spectra 8 6 3 RAW File 20 iuris a a a A ee tata 8 6 4 MakeBasis IN File 0 8 6 5 MEGA PRESS e 8 6 6 MakeBasis Diagnostics 8 6 7 Output BASIS File o a Calibrating Basis Spectra e 8 7 1 Calibrating GPC PCh amp NAA NAAG 8 7 2 Adding New Basis Spectra e SUMMATY o E a a aie da Shoe eto 8 CONTENTS 9 Further Useful Options and Information 102 9 1 Special Types of Spectra 2 o so 102 9 2 Muscle Spectra aco 2 E A A a a A 103 9 2 1 Standard Input sense ri dt la ees a 103 9 2 2 Additional Input e 104 9 23 Output ie see A Bea Ae eG Ae ia ce 105 9 3 Lipid Spectra Liver Breast Bone etc o 106 O
205. n permanently change the Archive Directory tree structure by clicking on the Reconfigure button in Fig 7 2 which produces a window like Fig 7 4 for Bruker Marconi Toshiba amp Varian data For these four data types a good alternative to their original default Archive Directory HOME 1cmodel saved is in the same directory as the original scanner data This is not the default for safety reasons only all original default output of LCModel LCMgui is confined to HOME 1cmodel The instructions in Fig 7 4 are quite detailed but the following step by step example may be useful 1 Click the green Append button next to Original Data Directory 2 Enter pdata lcmodel in the green field next to Enter a fixed element and then click the Append Entry button to the right This is the stage shown in Fig 7 4 3 If you are satisfied with the structure shown in the field next to Con structed Save Directory then click the Set Configuration button For example the LCModel output PS file will be tmp 1cm binary data bruker r0107 pdata lcmodel ps For Marconi Toshiba amp Varian data you might enter lcmodel rather than pdata lcmodel in Step 2 7 3 BASIC SETTINGS AND USAGE 63 Note the following You can experiment watching your progress in the field next to Con structed Save Directory and undoing mistakes by clicking the Start Over button You can cancel everything tha
206. n select a new Control Defaults File by clicking on its name which then appears in the green field e Click Use Selected File and LCMgui immediately repeats the conversion of your data and applies these new Control Defaults This newly selected Control Defaults File will also be used in the future until you change it again Note that the Control Defaults File only provides you with starting defaults You can then modify these values as you wish in the Control Parameters window 7 3 7 User Profiles There are two sets of default settings one for LCModel and one for LCMgui The Control Defaults file Sec 7 3 5 contains default Control Parameters for LCModel The gui defaults file contains default settings for LCMgui including the Control Defaults file path directories for archiving results and finding Basis Sets etc Both of these sets of defaults are saved in your User Profile If you have different types of data that require different LCMgui settings e g dif ferent Archive Directories then you should have a User Profile for each This would always be the case with data from different scanner types e g GE and Siemens Different users sharing the same account could also have different User Profiles You create a new User Profile by clicking on New Profile in Fig 6 6 which appears as you exit LCMgui The window in Fig 7 11 appears and you enter the name of the User Profile In Fig 7 11 I have entered toshiba
207. n should be supplied in the display command since trying to change these in the display program can sometimes crash the display program and LCMgui particularly if you are working over a slow network For example you might use gv orientation seascape With multi voxel data sets you can select a complete 2D slice with Fig 7 2 or a rectangular subset If your selection contains more than one voxel then Fig 7 13 will indicate this voxel range and will only offer a plot of the absolute value spectra in a 2D array as in PLOT 14 in HOME 1cmodel doc figures pdf This display following the DICOM standard may not always be consistent with the scanner s display Sec 11 4 3 2 If you want to save the plots they are in a file with a name of the form HOME 1cmodel temp 12d 21h 51m 09s 996pid plotraw PS where the numbers will be different as explained in Sec 7 5 2 2 Copy this to a safe place since the above directory will be deleted with the next analysis If no plot is produced check your plot and display command If a diagnostic message Aborting due to error Check User s Manual appears in your Unix window then check Sec 8 5 3 Check that your selected data file is correct Finally you could start LCModel which has extensive diagnostics 7 4 FURTHER USEFUL SETTINGS 71 zla Select Execution Script file 243 Below is a list of possible Execution Scripts which will be executed to run LCModel The selected
208. n white matter For human brain NAAG should be used The concentration must be at least 20 mM For rodent brain NAAG is often omitted At long TE do not use NAAG A shifted NAA is a good approximation Further more inconsistencies in concentration or signal strength between NAA amp NAAG are then fully excluded If you do use NAAG you should check the NAAG concentration by analyzing it as NAA as specified in Sec 8 7 1 Do not use Sigma NAAG it is often contaminated 8 2 1 3 Choline Containing Compounds Use GPC for the choline containing compounds In high quality spectra the advan tages of GPC over Cho can be seen Petra Pouwels personal communication 18 You should also use PCh in addition to GPC Unfortunately the usual forms of GPC CdCl adduct and PCh Ca salt cannot be used in the phosphate Standard Solvent in Sec 8 2 4 but be sure to use the formate and DSS markers with these two solutions as specified in Sec 8 2 4 The Cd also makes pH adjustment with NaOH impossible It would be better to find GPC without Cd but also without alcohols 8 2 2 Reference Markers Consistent referencing is essential Thus it is best to use the same referencing marker for all model spectra It should have no significant resonances in the Analysis Win dow T Michaelis uses 3 trimethylsilyl 1 propane sulfonic acid DSS also called TSPS Fluka no 92754 at a concentration say 1 mM high enough to precisely locate the
209. nce Basis Set singlet is not positive Hints Maybe you are integrating where there is no peak check your input of WSMET WSPPM or your BASIS file AREAWA manages the computation of the resonance area of the unsuppressed water signal Errors here cause the analysis to proceed without any scaling If you are attempt ing Water Scaling DOWS T then only the concentration ratios of the metabolites in the BASIS file will be valid If you are simulating metabolite Basis Spectra then the concentrations of all of the simulated metabolites would have to be mul tiplied by an unknown scale factor to be consistent with the concentrations of the BASIS metabolites If you are analyzing GE phased array data then the analysis will abort with FATAL AVERAG 5 w All AREAWA diagnostics are of type ERROR One of the following requirements has not been met NWSST gt 0 NWSEND lt NUNFIL NWSEND NWSST gt 9 The least squares analysis to estimate the water signal was too unstable Hints Never Happened but if it does try increasing NWSEND or check for gross errors in the H20 file such as all time domain points zero Same as AREAWA 2 The absolute value of one of the first NWSEND values in the H20 file is zero Hint Check for an error in FMTDAT of Namelist NMID of the H20 file that would cause zero values to be read AREAW2 Computes the resonance area of the unsuppressed water signal for Water Scaling 1 2 FATAL The water
210. nd PAGEHT must be positive 3 NUNFIL must be positive and must not exceed 65536 4 Same as INITIA 1 in Sec 12 2 5 Same as MYDATA 1 in Sec 12 2 6 Same as MYDATA 2 in Sec 12 2 7 You must have DEGPPM lt 9999 0 90 CHAPTER 8 MAKING THE BASIS SET 8 You must have SDPNTS lt 60 9 This would be my programming error 10 There is an error in your Namelist PLTRAW See Sec 12 1 1 on Namelist Errors 11 End of File reading your Namelist PLTRAW See Sec 12 1 1 12 There is an error in FMTDAT in your Namelist NMID in your RAW file or the format of your time domain data in your RAW file does not match FMTDAT See Sec 12 1 1 on Namelist Errors 13 End of File reading your RAW file See Sec 12 1 1 14 You must have PTTITL amp PTLABL positive and RHTITL amp RHLABL at least 1 0 15 Same as ZEROVX 4 PLOT2 1 Same as PLTRAW 14 2 Same as ONEPAG 4 in Sec 12 2 3 Same as ONEPAG 8 in Sec 12 2 AXRND1 1 One curve to be plotted consists of only one point Are PPMST and PPMEND very close to each other 2 All 3 elements of NAIM must exceed 1 One curve to be plotted consists of at most two points Are PPMST and PPMEND very close to each other w 8 6 Running MakeBasis You run MakeBasis with a command like HOME 1cmodel bin makebasis lt makebasis in where makebasis in Sec 8 6 4 contains the changes to Control Parameters An example is HOME 1cmodel test control makebasis in used for the test run in Chap 4 In addition a
211. ng you could change to IAVERG 3 Sec 11 9 2 but Water Scaling does no harm since it does not change the concentration ratios 146 CHAPTER 11 FINE POINTS 11 9 2 Other Spectra with Identical Phases amp Referencing Shift Use Sec 11 9 1 for GE amp Toshiba data Use the Siemens utility to combine their multi channel data For others this is usually multi channel data from a phased array or a Philips MEGA PRESS file see Sec 9 4 1 2 Rarely it is just a time series of spectra with identical phases and referencing With phased array data in order to coherently sum each channel you must do eddy current correction Sec 6 3 3 For this you must have a reference data set from the same voxel with no water suppression You then use IAVERG 1 defined below You must put the data sets in the format of CSI data sets specified in Sec 5 2 3 1 With phased array data Varian automatically does this for you Philips does this for you producing a file as a series of spectra usually in a row e g in a series of scans of the same voxel For Philips MEGA PRESS see Sec 9 4 1 2 You then simply select this file with LCMgui with phased array data activate eddy current correction and input one of the following values for TAVERG IAVERG INTEGER IAVERG 1 to analyze a weighted average of the spectra Use this if you also have a corresponding set of unsuppressed water reference spectra from the same voxel This is required for the statistical w
212. nother Preprocessor now and in the future then click on its name and its name will also appear in the green field as the Selected File Then click Use Selected File and the selected Preprocessor would be executed immediately and in the future until you change it GE Preprocessors The following Preprocessors are available for GE data change phase priors GE automated Probe spectra usually need only very small phase corrections and this prior information is used by default However this can cause huge phasing errors if you have non standard data requir ing large phase corrections If analyses of non standard data show poor phasing then you should first edit the file HOME lcmodel1 preprocessors change phase priors as explained in the file where ge5 or gelx Then select this Preprocessor in Fig 7 15 select frames can salvage a spectrum that has been damaged by patient mo tion You can select a subset of frames that excludes the period where there was motion or other damage Full instructions are in the com ments in HOME 1cmodel gelx preprocessors select frames Note that LCMgui must not be running in background i e it must not have been started with a amp at the end of your command select frames is not available for 5 x data or for multi channel data set 3t bandwidth LCMgui assumes that the bandwidth of 3T data is 5000Hz If this is not the case with your 3T data then edit th
213. nus the fit to the data The Residuals are a sensitive diagnostic of the analysis In PLOT 1 in HOME 1cmodel doc figures pdf they look fairly randomly scattered about zero indicating a fit to within experimental error with no apparent systematic errors Section 3 5 gives a series of examples and guidelines on the diagnostic use of the Residuals and Baseline 2 3 Diagnostics The table below the Concentration Table summarizes all the diagnostic messages from LCModel during this analysis Always look at this table The 5 types of diagnostics in increasing order of severity are info warning boldface blue and all in boldface red ERROR FATAL and ILLOGICAL Chapter 12 lists and explains each diagnostic and suggests possible remedies You should consult Chap 12 if ERROR or warning occurs The type info is generally harmless When FATAL or ILLOGICAL occurs the run aborts and you must consult Chap 12 If the run aborts whatever is available is plotted e the usual fit to the data usually not available or e the absolute value of the frequency domain data the Absolute Value Plot or e no plots only the tables of diagnostics and input changes if the raw data have not yet been read in 18 CHAPTER 2 ONE PAGE OUTPUT The Absolute Value Plot of the test data is illustrated in PLOT 5 in HOME 1cmodel doc figures pdf where the FATAL was caused by inputting un reasonable values for a Control Parameter DOREFS The most c
214. o be detectable in your spectrum then you should exclude these from this analysis as follows NOMIT INTEGER the number of metabolites to be excluded You can exclude Basis Spectra in the BASIS file as well as simulated spectra Default NOMIT 0 CHOMIT CHARACTER 100 6 the first NOMIT elements contain the Metabolite Names of the spectra to be excluded For example some pathologies have Glyc much stronger than Ins at least at long TE Sec 8 2 1 1 You add a simulated Gly by inputting Sec 11 7 NSIMUL 14 and you may want to omit Ins Sec 8 2 1 1 with NOMIT 1 CHOMIT 1 Ins CHCALI amp NCALIB Sec 8 7 1 are much more convenient to use when you want to omit all but one or a few of the Basis Spectra e g for calibration analyses 9 8 2 Keeping Basis Spectra in the Analysis When PPMST or PPMEND restricts the Analysis Window LCModel automatically omits from the analysis metabolites that are weakly represented in the Analysis Window e g Lac amp Ala in PLOT 16 in HOME 1cmodel doc figures pdf You can force some of these to be included as follows 118 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION NKEEP INTEGER the number of metabolites that are to be kept in the current analysis regardless of the ppm range of the Analysis Window You can include Basis Spectra in the BASIS file as well as simulated spectra Default NKEEP 0 CHKEEP CHARACTER 300 6 the first NKEEP elements contain the Meta
215. o omit them from the analysis The weak priors allow for the variability of these but still help to avoid unrealistic concentrations in low information spectra This still introduces a weak bias but not as strong as omitting them which fixes them to zero Weak bias toward their expect values should yield more accurate overall analyses than setting them to zero or giving them complete freedom to assume huge values If e g in a pathology one of these metabolites is unexpectedly well determined and strongly deviates from expect by more than four times stddev then the line for metnum in the One Page Output is in bold red italic with a plus or minus preceding the Metabolite Name Sec 9 9 1 3 If this deviation seems plausible then you should relax or remove this prior as specified in Sec 11 8 1 For example with rodents Tau is often high and you should probably eliminate its prior as illustrated in Sec 11 8 1 11 9 COHERENT DATA AVERAGING 145 In a restricted study with more accurate expect from controls you might be able to strengthen some constraints However do not make stddev expect too small Their main purpose is to bring stability not self deceptive bias Very small values would also increase the difficulty in the constrained nonlinear optimization algorithm in LCModel 11 8 1 Omitting Ratio Priors If you want to eliminate a few Concentration Ratio Priors then the following are usually more convenient than rearranging t
216. o set IPAGE2 0 to get only one page per analysis Use a PostScript viewer like gv or evince 11 4 6 Skipping bad voxels With LCMgui and LCModel version 6 3 0D and above LCModel automatically skips to the next voxel when a voxel causes an abort in a multi voxel analysis It will do this up to MAX_BAD 16 times before stopping If necessary you can change MAX_BAD in the Execution Scripts standard or nice However if there is an abort with the very first voxel LCModel stops so that you can check your input for gross errors The settings below are no longer needed with version 6 3 0D and above You can skip over voxels as follows NVOXSK INTEGER the number of voxels to be skipped Default NVOXSK 0 ICOLSK INTEGER 64 ICOLSK J specifies the column of the J th voxel to be skipped IROWSK INTEGER 64 IROWSK J specifies the row of the J th voxel to be skipped 140 CHAPTER 11 FINE POINTS 11 5 Basis Spectra in the Preliminary Analysis The Preliminary Analysis for initial referencing and phasing uses a reduced set of Basis Spectra that are meant to contain only the major landmarks in the spectrum being ana lyzed Since crude Gaussian Lorentzian lineshapes are used in this Preliminary Analysis weak metabolites cannot be estimated anyway The Basis Spectra used for this Preliminary Analysis are specified as follows NUSE1 INTEGER the number of Basis Spectra that are to be used in the Preliminary Analysis Default NUSE1 5
217. of the CHCALI Thus you would be calibrating a Basis Spectrum with itself Not used Not used FATAL When the simulated Basis Spectra are added the total number of Basis Spectra to be used exceeds 100 Hints Check and reduce NSIMUL If you really want to use so many then I can send a new version FATAL A CHSIMU cannot have more than 20 simulated peaks i e 20 ppm in Sec 11 7 FATAL The elements of all CHSIMU Sec 11 7 must satisfy fwmins lt fwez and sdfw gt 0 warning The elements fwmin amp fwex of a CHSIMU Sec 11 7 are too close 12 2 STANDARD LCMODEL DIAGNOSTICS 163 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 together To avoid numerical problems fwex will be slightly increased warning The element sdfw of a CHSIMU Sec 11 7 is too small To avoid numerical problems sdfw will be slightly increased warning The element sdppm of a CHSIMU Sec 11 7 is too small To avoid numerical problems it will be increased slightly FATAL sdppm lt 0 FATAL The elements of all CHSIMU Sec 11 7 must satisfy ampx 4 0 and Ippm lt 998 Hints Check the format of your CHSIMU for erroneous or blank entries Not used FATAL The elements fwmin of all CHSIMU Sec 11 7 must satisfy fuwuminx gt 0 Same as MYBASI 16 FATAL Incorrect values for the Hidden Control Parameters RINCSH SDSHMN or SDSHMX Do not input these FA
218. ols below The tokens and rules in Sec 11 7 must be followed exactly The errors here refer to wrong tokens or format The actual values are checked elsewhere The number of spaces between elements should not be important but only spaces not tabs etc are allowed O The end of CHSIMU could not be found Are there more than 20 peaks i e more than 20 ppm Is CHSIMU more than 528 characters It must end with a number preceded by AMP nn Error reading metabo Inn Error reading ppmi 2nn Error reading sdppm FWHM 3nn Error reading fwmini lt 4nn Error reading fwer 5nn Error reading sdfw AMP 6nn Error reading amp or end of CHSIMU instead of Tnn Error reading ppm FWHM where gt 1 8nn Error reading fwmin AMP where x gt 1 9nn Error reading amp or end of CHSIMU instead of where gt 1 PARSUM parses the Concentration Sums metdenom amp metwt in Sec 11 8 At least part of the illegal string is output at the end of the Detailed Output All PARSUM diagnostics are FATAL 12 2 STANDARD LCMODEL DIAGNOSTICS 167 1 A metdenom or metwt has more than 211 characters There is probably an error in the CHRATO string 2 One of the Metabolite Names in the Concentration Sum has more than 6 characters 3 A Concentration Sum has more than 30 Metabolite Names PHALIP Phases lipid spectra 1 The Hidden Control Parameter PPMST_PHALIP lt PPMEND_PHALIP Do not input these parameters PHASTA Gets starting estimates for
219. omly scattered about zero as in PLOT 1 in HOME 1cmodel doc figures pdf Reject an analysis with strongly non random residuals as in PLOT 15 with system atic peaks and valleys below 1 8 ppm indicating an inadequate model or artifacts in the data 3 5 2 Wild Baseline The Baseline must be flexible enough to account for a wide variety of unpredictable complications including 1 artifacts in the data 2 substances not present in the Basis Set of model metabolite spectra 3 substances possibly in the Basis Set but with abnormally short T2 times 4 inaccuracies in the simulated models for the highly variable lipid and macro molecule signals 5 incomplete water suppression Item 1 often leads to failure as in PLOT 15 in HOME 1cmode1 doc figures pdf where wild Baseline oscillations are not enough to fit the artifacts PLOT 17 is an analysis of the data from PLOT 1 It was forced to fail by inputting a grossly incorrect constraint on the referencing shift The Baseline has huge dips and huge Glu Ins concentrations trying to fit peaks of other metabolites displaced by the huge referencing error Here the model is wrong with Glu trying to fit NAA etc and the small SD values for Glu amp Ins are therefore also wrong Reject analyses with wild Baselines Huge dips down away from the data as in PLOT 17 are particularly bad with huge concentrations being assigned to metabo lites trying to fit the data with the
220. ommon causes of a FATAL are input errors which are usually obvious from the diagnostics or hopelessly bad data which is often immediately apparent from the Absolute Value Plot 2 4 Miscellaneous Output The lines in this table are 1 FWHM full width at half maximum is a rough estimate of the linewidth in the in vivo spectrum S N is defined here as the ratio of the maximum in the spectrum minus Baseline over the Analysis Window to twice the rms Residuals This seems to be widely used 6 p 218 but some may use twice this value This S N is also rough because it depends on offsets in the model spectra and Baseline Neither of these single numbers can unambiguously characterize the quality of your data They are only rough guides Only the SD values in the Concentration Table simultaneously account for both resolution and noise level as well as many other things to give the most useful guides to the reliability of the estimates 2 Data shift the referencing shift performed on the data spectrum 3 Ph the zero order and first order data phase corrections dy and 1 using the conventions in the definition of DEGPPM in Sec 8 5 2 Section 9 9 4 explains optional additional output that you can easily get in this table 2 5 Input Changes This last table lists all input that was made to change default Control Parame ters in LCModel These are usually automatically supplied by LCMgui or the Pre Formatting Program to set
221. on ECC in ParaVision only with single voxel data Do not do ECC again with LCMgui For Water Scaling you should be able to use the file fid refscan as your unsuppressed water reference file It is in the same directory as your water suppressed file fid Only if fid refscan is absent use fid ref but only if it has the same size as fid A reference scan fid file is absent in PV6 0 use PV6 0 1 or higher instead If you use LCMgui see Sec 7 3 1 If you do not use LCMgui you will probably have to first use Bruker s convdta to correct fid for Bruker s digital filter 3 7 2 GE Probe raw P files are highly recommended They are originally in the directory usr g mrraw with Signa 5 2 5 7 filenames of the form Pnnnnn and with Signa LX and higher filenames of the form Pnnnnn 7 Single voxel processing with LCMgui is then pretty much automatic including phased array data LCMgui can also handle some spectrum G files Signa 5 x and old spec tro raw P files LX and higher but these are not as complete as Probe raw P files LCMgui only handles single voxel GE files For CSI data sets use the Re search Version of SAGE which now has Mary McLean s LCModel interface 9 To use this SAGE LCModel interface you must read Secs 11 4 1 amp 7 2 5 3 It is best to have a separate user account and installation for each vendor GE Probe data require very little phase correction and the Control Pa rameters SDDEGP amp SDDEG
222. on is weak and the unsuppressed water reference not very good you may see that ECC obviously amplifies the noise in the high ppm part of the spectrum If this amplified noise extends into the Analysis Window then you should switch off ECC by removing DOECC and SDDEGZ from the CONTROL file for this spectrum only 5 3 CONTROL FILE 41 5 3 5 First Order Phase Correction The first order phase correction 61 is often small and prior knowledge of this can be used to speed up and improve analyses At least two well separated landmark peaks are needed to determine the zero order and first order phase corrections Therefore prior knowledge of phases is very important with spectra lacking two strong land marks as in PLOT 6 in HOME 1cmodel1 doc figures pdf although this case also worked without specifying SDDEGZ or SDDEGP As specified in Sec 9 7 the Control Parameter SDDEGP tells LCModel the standard deviation of the expectation value DEGPPM of the first order phase correction The following seem to work well for single voxel data e GE Probe data SDDEGP 1 e Marconi Picker data with STEAM and current PRESS sequences SDDEGP 2 DEGPPM 2 5 e Marconi Picker data with very old PRESS sequences SDDEGP 2 DEGPPM 28 If you are not sure whether your PRESS sequence is older or newer then try both 28 and 2 5 for DEGPPM they are very different and the wrong one should give very poor phasing If you are in doubt s
223. on with very noisy data or very smooth lineshapes 7 warning ag has reached its minimum allowable value Hints This can happen with very narrow lineshapes e g narrower than the Basis Spectra as often in vitro 8 info ag and ag have both reached their maximum allowable values Hints This is common with very noisy data or very smooth lineshapes and Baselines 9 warning ag and ag have both reached their minimum allowable values Hints Working Hard 10 ERROR A nonlinear least squares analysis has failed corrective action will be attempted Hints Never Happened 11 info A local minimum in the objective function has been detected and corrected Hints Working Hard 12 info A local minimum in the objective function in an unregularized anal ysis has been detected and corrected Hints Working Hard 13 info Numerical instabilities have caused the objective function to increase corrective action is taken Hints Working Hard 14 warning Maximum iterations in looking for an acceptable regularizor Hints Working Hard 15 FATAL All nonlinear least squares analyses in all iterations failed Hints Never Happened but if it does check for very poor or erroneous data 16 ERROR Somehow the objective function or the number of free parameters has gone to zero Corrective action will be attempted Hints Never 12 2 STANDARD LCMODEL DIAGNOSTICS 169 17 18 Happened warning A regularizor satisfying the criterion i
224. ontrol Parameter PRMNMX which you should not change It could also be caused by a perfect fit to the data or less than one effective free parameter in the full model in 1 Eq 1 for which the hint of FINOUT 6 applies It could also be caused by more free parameters than data points in the Analysis Window for which the Hints of FISHNI 1 apply ERROR There are less than two free parameters in the full model in 1 Eq 1 and the approximation for the sum of squares could be poor This could lead to bad regularization parameters FATAL The sum of squares would be so large that it would cause an arith metic overflow Hints See the discussion of overflows in Sec 12 1 2 warning The solution of a nonlinear equation is inaccurate This could in rare cases cause bad regularization parameters ERROR Same as FSHSSQ 4 ERROR Same as FSHSSQ 4 FTDATA Fourier transforms the time domain data and shifts the spectrum 1 FATAL The sum of the absolute values of all the points in the Analysis Window is zero Hint a Your RAW file is probably being read in as all zeroes or extremely small values check the format of the data in this file b This will also occur if you are using BASCAL T and your CHBCAL specifies a Metabolite Name that is not in the Basis Set GBACKG Get background computes the B spline basis set for the Baseline 1 2 3 4 Not used FATAL There is not enough storage for NBCKMN spline knots Hints
225. or GPC PCh than for GPC and PCh CrCH2 is a correction term rather than a metabolite of interest It is a simulated negative creatine CHo singlet around 3 94 ppm This corrects for attenuation of the Cr CHo singlet due to water suppression and differential relaxation effects at longer TE Gua is a simulated singlet to account for occasional significant signals at 3 78 ppm especially in tumors The Gua stood for guanidinoacetate 4 but in nearly all cases this assignment is probably wrong and Gua is only an empirical fit to the signal at 3 78 ppm which may be partly due to GSH 2 1 2 Lower Part The lower part of the Concentration Table contains simulated lipid Lip and macro molecule MM components As with CrCH2 they are often considered nuisance correction parameters but they may also be of interest e g in classifying tumors 5 Section 11 7 defines the Metabolite Names in this lower part It also explains how to specify your own simulated components The numbers after MM for macro molecules and Lip for lipids indicate the approximate chemical shift in ppm of the peaks e g MM14 for the macromolecule peak near 1 4 ppm which may also contain lipid contributions If Water Scaling Sec 6 3 4 is used to estimate the concentrations in the upper part in mM then all of the MM amp Lip will be in mM of CH2 groups except for Lip09 amp MM09 which will be in mM of CH3 groups Otherwise the unknown sc
226. orted or poorly defined e g as sometimes with digital water suppression or in diffusion experiments and if your spectra usually have well defined landmarks e g NAA Cr Cho etc then you should avoid using the water peak by leaving DOREFS 1 F On the other hand if you are using weak water suppression with a well defined water peak then it is best to set DOREFS 1 T if in addition the climb toward the residual water peak is strong enough so that the Baseline at 4 0 ppm is comparable to the Cr Cho amp NAA peaks then you should also set DOREFS 2 F Plotting the absolute value of a few spectra with Preview Data in LCMgui Sec 7 4 1 can help you make this decision Default DOREFS F T i e water is not used and the usual metabolite landmarks are used NREFPK INTEGER 2 NREFPK JCCF is the number of 6 functions reference peaks for CCF JCCF It cannot exceed 10 Default NREFPK 1 1 for water and NREFPK 2 3 for the usual landmarks listed below PPMREF HZREF REAL 10 2 the position of 6 function JREF for CCF JCCF is PPMREF JREF JCCF HZREF JREF JCCF where PPMREF is in ppm and HZREF is input in Hz Defaults PPMREF 1 1 4 65 for water PPMREF JREF 2 2 01 3 03 3 22 for NAA Cr amp Cho all HZREF elements zero 11 4 MULTI VOXEL DATA SETS 137 To use the Lac doublet input two PPMREF 1 33 and the corresponding two HZREF 3 6 amp 3 6 However lipid signals can be mistaken
227. ou must also multiply the LCModel concentrations by the correction factor for TR which for STEAM is fon A exp rm 1 __1 1 z exp RDvivo T Lvivo T Ivivo T lyitro where RD TR TM TE 2 or to good approximation for both PRESS amp STEAM i f exp T Ryivo T Lvivo since we require T Rvitro gt Tl1vitro for the basis set This correction factor usually becomes unimportant with T Ruivo gt 5000 ms frr 11 3 Prior Referencing Information Spectra without the usual landmark peaks of NAA Cr amp Cho can be difficult to refer ence automatically The settings in Sec 9 1 can often help If not in case of a referenc ing failure which often looks like PLOT 17 in HOME 1cmodel1 doc figures pdf you can often salvage the spectrum by helping LCModel with more specific referenc ing information Try the following in the order listed 11 3 1 Off Resonance Spectra If the center frequency of your spectra is further away from the water resonance than expected by default about 0 1 or 0 2 ppm and you sometimes get results like PLOT 17 in HOME 1cmodel doc figures pdf use PPMCEN REAL is the ppm value at the center of the spectrum to within about 0 1 ppm Default PPMCEN 4 65 It may be absolutely necessary to input PPMCEN if your spectral bandwidth is so narrow and the frequency offset so large that the Analysis Window would otherwise appear to be at the edge of your spectrum Note that if you do eddy cur
228. ould try referencing to Cho by not inputting DOREFS and instead inputting PPMREF 1 2 3 2 Instead of the above you could reference to the maximum over the whole range of 3 8 ppm to 1 0 ppm This is often EMCL at 1 52 ppm You then input PPMREF 1 2 1 52 Instead of the above you can specify the referencing shift with PPMSHF Sec 11 3 3 9 2 2 6 Phasing Phasing can also be difficult without a strong Cr peak If you have prior phasing in formation you should definitely use it Sec 9 7 with SDDEGP This is all automatically done with single voxel GE Picker amp Toshiba data 9 2 3 Output See PLOT 8 in HOME 1cmodel doc figures pdf With Water Scaling the Conc column of the Concentration Table has the ratio of the metabolite resonance area to the unsuppressed water resonance area The term resonance area is used instead of peak area since LCModel attempts to resolve overlapping signals even when they form only one combined peak Without Water Scaling only resonance area ratios such as IMCL13 Cr are meaning ful The resonance areas of the following signals are in the Concentration Table IMCL13 the IMCL CH signal around 1 3 ppm EMCL15 the EMCL CH signal around 1 5 ppm 106 Cr I09 E11 Water 153 E55 121 E23 tau cr39 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION the Cr CH signal around 3 03 ppm However some of the orientation dependent Tau signal may also con
229. p control LCMODL LPS 0 LPRINT 6 FILPRI dump print PPMCEN 999 END Run LCModel HOME 1cmodel bin lcmodel lt dump control The changes to LPRINT and FILPRI produce dump print the Detailed Output file The unreasonable value for PPMCEN produces a FATAL diagnostic and abort almost immediately but not before the complete Namelist LCMODL has been written in dump print All Control Parameters are in Namelist LCMODL and can therefore be modified by the Namelist input in the CONTROL file input to LCModel Since dump control only modified LPS LPRINT FILPRI and PPMCEN all other values output in dump print are the original default values The original defaults for the modified four are LPS 8 LPRINT 0 FILPRI PPMCEN 4 65 This Namelist output is ordered in four groups according to Fortran type 11 16 ANALYZING MAGNITUDE SPECTRA 151 CHARACTER INTEGER LOGICAL amp REAL with alphabetical ordering within each group Most of these Control Parameters need not be changed or even checked Many are so called Hidden Control Parameters which are not documented in this manual and should not be changed These are already set to their optimal values 11 16 Analyzing Magnitude Spectra This should only be done if it is impossible for LCModel to phase your spectra For example some CSI sequences have huge unknown 1st order phase corrections and it is necessary to analyze the absolute values of the spectra
230. put You get one PostScript output file for each voxel If your default display print com mand is a display command then you will get a series of displays that each have to be closed before you view the next It is usually more convenient to suppress the display print command by deleting it using Sec 7 4 6 You can then use Sec 11 4 5 to combine all the files into one for convenient viewing 6 3 6 The Rest of the Control Parameters Window The rest are briefly listed here with references to more detailed descriptions They make LCMgui very flexible TITLE This field at the top of Fig 6 4 will also appear at the top of the One Page Output You can edit this field e g to insert comments on location or pathology Save File Types See Sec 6 5 Advanced Settings Clicking this button produces a menu Fig 7 6 offering several new windows In this menu the View Edit Control Parameters window allows you to view change delete or add Control Parameters For example if the ECC button in Sec 6 3 3 was off and you turned it on blue then you should check the Control Parameter SDDEGZ as specified in Sec 7 3 4 Preview Data This allows you to first view your data in the time and or frequency domain on your screen or printer e g to make sure that you have the correct data and that it is worth analyzing Sec 7 4 1 6 3 7 Reload Data The TITLE line or Preview Data may cause you to wonder whether you selected the correct da
231. r ge5 gelx other marconi philips siemens toshiba or varian You can always examine these in their directories and delete the obsolete ones You can also be careful with rm fr delete a whole User Profile with a command like rm fr HOME lcmodel profiles my profile 7 5 2 2 Temporary Files These are automatically purged and you normally do not have to bother with them Each LCModel analysis typically requires only about 200 kB of temporary disk stor age more with CSI data They are all in HOME 1lcmode1 temp Each time LCMgui prepares data for an LCModel analysis a unique temporary directory is created e g HOME 1cmodel temp 12d 21h 51m 09s 996pid The numbers are respectively the day of the month hour minute second and process ID When the LCModel analysis is finished the directory is marked for deletion and is deleted when the next LCModel analysis is prepared So directories only remain for LCModel analyses not completed including aborts from LCMgui By default these remaining directories are deleted if they are more than 86400 0 s 1 day old You can modify this value in the following line in gui defaults i_deleteAfterSeconds 86400 0 Chapter 8 Making the Basis Set It is only necessary to read this chapter if you must collect your own Basis Set of in vitro model metabolite spectra Your in vivo data must be compatible with the Basis Set as specified in Sec 8 1 If you have been supplied with a
232. r data e g naa to make usr myname press25 naa Then your ECC data will go into the file usr myname press25 naa ecc Click on Change Execution Script in the Advanced Settings menu select ecc in the window in Fig 7 14 and then click Use Selected File By default your LCMgui print command will output plots of your ECC data HOME 1cmodel execution scripts ecc shows you how to change which plots will be output and how to suppress all plots although plots of the absolute value of the spectrum are recommended You can also suppress all plots by clicking on Change LCMgui Settings in the Advanced Set tings menu and then changing the Print Command to touch Now you can click Run LCModel Unless you have Marconi Picker data or GE Probe P files you will then be instructed to select the corresponding unsuppressed water reference file You may be asked to select a BASIS file You can select any file it will not be used Click on Save Changes in Fig 7 8 In Fig 7 9 enter a new name e g ecc next to New filename and click Use Next Time Then the above changes will automatically be used for ECCs on your other files during this 86 CHAPTER 8 MAKING THE BASIS SET LCMgui session 12 Click on Next Analysis to prepare an ECC of your next data file For further ECCs during this LCMgui session do the following e Step 1 above Select your water suppressed data
233. r of these conventions LCModel needs three input files RAW for the LCModel RAW file with time domain data defined in Sec 5 2 GE amp Siemens raw files will be referred to in lower case BASIS for a file containing the Basis Set of model metabolite Basis Spectra de fined in Chap 8 33 34 CHAPTER 5 RUNNING LCMODEL WITHOUT LCMGUI BASIC INPUT CONTROL for the file with the changes to the Control Parameters input to LCModel defined in Sec 5 3 For a given acquisition protocol and field strength the BASIS file is generally the same whereas the RAW and CONTROL files must be prepared for each spectrum Some other extensions used in this manual are PS for a PostScript file that you can display or send to your PostScript printer IN for a file with analogs to Control Parameters input to the MakeBasis PlotRaw or KECC programs in the LCModel package 5 1 2 Control Parameter Conventions When a Control Parameter is defined in this manual its name 6 or fewer charac ters is followed by its Fortran type in parenthesis Some examples of Fortran type specifications are CHARACTER 56 means a string of a maximum of 56 characters enclosed in single quotes as NAMREL Cr Note that the string Cr is case sensitive but the Namelist name NAMREL is not Thus you could also use namrel Cr but not NAMREL cr INTEGER 20 means an integer array with up to 20 elements as for KEY in Sec 5 3
234. r spreadsheets As specified in Sec 11 4 3 each voxel has its own set of files Only the CSV file combines the concentration data from all voxels into a single file which can be imported to spreadsheet programs With LCMgui this file is produced automatically in your output directory It has the exten sion csv and is usually named spreadsheet csv Without LCMgui you specify it using Sec 5 3 7 2 Often spreadsheet programs recognize the csv extension and import it automatically Each wide row contains all of the concentration estimates for one voxel The first two columns contain the row and column numbers of the original multi voxel grid With Excel OpenOf fice org amp LibreOffice you can easily sort these in ascending order You can also select the table and then use the Format Column Autofit Selection or Optimal Width option to improve the column widths In the Data menu you can use Pivot Table Excel or Data Pilot to easily produce the 2D grid of e g NAA NAAG values 11 4 5 Combining Multi Voxel PS files You can combine all the files with names ending with ps in a directory e g my dir into a single file with the following commands cd my dir HOME 1cmode1 bin cat paged ps This produces the PostScript file my dir cat paged ps which you can comfortably page through with your PostScript viewer although the ordering may not be optimal with large subsets e g with Row 10 coming before Row 2 You may want t
235. ra for a Basis Set are acquired under identical conditions including coil loading so that the effective gain is the same for all Otherwise you must scale them as specified in Sec 10 1 1 However the spectra should still be acquired under the same conditions voxel in the center of the phantom phantom position in the magnet etc as well as the more 84 CHAPTER 8 MAKING THE BASIS SET obvious conditions in the protocol It is therefore best if all spectra are acquired in the same time period If you nevertheless later want to add some new Basis Spectra to an old Basis Set then e If both old and new were acquired with formate and DSS markers as in Sec 8 2 4 then you can simply process them all together using Auto Scaling with MakeBasis You need only specify the different concentrations of the formate with CONCSC e Otherwise you must also acquire one new spectrum of a metabolite that is also in the old Basis Set and scale these two together as specified in Sec 8 7 2 8 3 3 Acquisition Time and Bandwidth These parameters must be the same for each spectrum in the Basis Set However they do not have to be the same as in your in vivo data The acquisition time for one scan must be long enough so that the time domain data decay into the noise Otherwise oscillations due to truncation effects in the DSS and formate markers can hurt the automatic scaling and phasing Particularly at higher fields you may need to use a longer acq
236. raging Concentrations over a Series of Analyses 11 2 Relaxation Corrections o o ooa a ee 11 3 Prior Referencing Information 2 04 11 3 1 Off Resonance Spectra 2 ee 11 3 2 Referencing to Water piou aiy k dor aai n 02000 eee eee 11 33 Inputting a Starting Shift 11 3 4 Fixing the Shift s acr a di eses as ee 11 3 5 Details of Automated Referencing 11 4 Multi Voxel Data Sets 2 0 0 20 200 0 020000 2000 11 4 1 GE CSI data 0 200 002 0002 0 00 11 42 Non GE CSI data e 0 02020 000 11 4 3 Multi Voxel Filenames o a 11 4 4 Output for spreadsheets o o 11 4 5 Combining Multi Voxel PS files 0 11 4 6 Skipping bad voxels o 0 020000 11 5 Basis Spectra in the Preliminary Analysis 11 6 Unusual Phantoms a ss masia 2 4 G03 e Pa ee ae ee ee 11 7 Simulating Basis Spectra ee WT PAs Methodi a0 atch ete Dey oe tee PO ie ek hE ang LP 11 8 Concentration Ratio Priors 02000022 ee 11 8 1 Omitting Ratio Priors o e 11 9 Coherent Data Averaging 2 2000002 eee 11 9 1 GE Toshiba Phased Array Single Voxel Data 11 9 2 Other Spectra with Identical Phases amp Referencing Shift 11 9 3 Varying Phases or Referencing Shifts 11 10 Analyzing a Time Series of Spectra 0
237. re is not enough room on the first page IPAGE2 2 to always output a second page with the tables 5 3 7 Archiving the Results The PS and TABLE files contain the essential graphic and numeric information from the analysis The CSV file contains the concentrations and can be imported to spreadsheet programs With LCMgui you can select any of these to be output using Fig 7 3 5 3 7 1 TABLE File The TABLE file contains all the information in the tables of the One Page Output as a self documented text ASCII file You can archive these files particularly if you have assigned them unique pathnames with FILTAB into a database e g for computing means or other statistics over a group of spectra For this you must input the following FILTAB CHARACTER 255 table file the pathname of the TABLE file to be output e g home user1 1cmodel saved pat_777_77 table LTABLE INTEGER LTABLE 0 the default will suppress creation of this file LTABLE 7 will make this file 5 3 7 2 CSV Spreadsheet File The CSV file for spreadsheets is convenient for multi voxel analyses Sec 11 4 4 For single voxel analyses it consists of only two very wide rows of concentration data FILCSV CHARACTER 255 the name of the CSV file to be output e g gt home user1 1cmodel saved pat_777_77 spreadsheet csv A file with the extension csv but not a file simply with the name csv is automatically imported as a co
238. reis B D Ross Absolute quantitation of water and metabolites in the human brain I Com partments and water J Magn Reson B 102 1 8 1993 178 29 30 33 34 BIBLIOGRAPHY P Vermathen C Boesch R Kreis Mapping fiber orientation in human muscle by proton MR spectroscopic imag ing Magn Reson Med 49 424 432 2003 C Boesch J Slotboom H Hoppeler R Kreis In Vivo determination of intra myocellular lipids in human muscle by means of localized H MR spectroscopy Magn Reson Med 37 484 493 1997 W C Hamilton Statistics in Physical Science Ronald New York 1964 p 42 or another textbook on elementary applied statistics J R Alger A J Frew T F Cloughesy W del Vecchio J P Villablanca J G Curran Novel methodology for the archiving and interactive reading of clinical mag netic resonance spectroscopic imaging Magn Reson Med 48 411 418 2002 M A McLean R J Simister G J Barker J S Duncan Metabolite nulling improves reliability of LCModel analysis of short echo time spectroscopy Proc Intl Soc Magn Reson Med 10 529 2002 P G Henry G Oz S Provencher R Gruetter Toward dynamic isotopomer analysis in the rat brain in vivo automatic quan titation of 1 C NMR spectra using LCModel NMR Biomed 16 400 412 2003 Index absolute concentrations 124 LCMgui 48 absolute value analyses 151 Absolute Value Plot 17 ABSVAL 151 acquisition para
239. rent correction Sec 6 3 3 or 9 6 you normally must not use PPMCEN the referencing shift is usually corrected by ECC This assumes that the unsuppressed water reference spectrum has about the same shift as the water suppressed spectrum If the water reference spectrum is not shifted then you cannot use PPMCEN you can only use PPMSHF Sec 11 3 3 11 3 PRIOR REFERENCING INFORMATION 135 11 3 2 Referencing to Water If your water suppression is not too extreme it is convenient to use the huge water peak as an approximate referencing landmark You can check this by plotting several of your spectra with PlotRaw LCMgui users see Sec 7 4 1 to see if there is a reasonable water peak around 4 65 ppm The water peak need not be exactly at 4 65 ppm and it can be distorted since it only tells LCModel where to start looking But it cannot have been totally removed e g by some digital filter there must always be a clear maximum in the correct region If a reasonable water peak is present then you can use this at least in problem cases for referencing by setting the Control Parameter DOREFS 1 T You may also have to extend the maximum expected shift with SHIFMN 1 or SHIFMX 1 in Sec 11 3 5 This could be a universal solution for all of your analyses where you automatically input DOREFS 1 and SHIFMN 1 or SHIFMX 1 in every CONTROL file 11 3 3 Inputting a Starting Shift Less automatic but more direct is to plot the absolute valu
240. repeat the changes every session The selected Control Defaults file if any that has been used so far in this session is shown in green right below Selected File You can select another file by clicking on its name below Click on Do Not use any file below to use no Control Defaults file at all Click on Use Selected File below when you are satisfied with the green file An empty green field means that no file will be used You can later check and edit the Control Parameters resulting from your selection in the View Edit Control Parameters window in the Advanced Settings menu The files can be viewed in their directory SP sp lcmodel profiles 1 control defaults Selected File bascal gpc pch bm10 bm14 bm14 tumor bnz bn2 tumor Use Selected File Do Not use any file Figure 7 10 Here you select the Control Defaults file to be used now and in the future Defaults File used at the start of this session will again be used 7 3 6 Changing your Default Control Parameters How do you know what file Control Defaults file is currently being used You again click on Advanced Settings in Fig 7 2 The Control Defaults file currently being used is in parentheses next to Change Control Defaults file In Fig 7 6 there is nothing between the parentheses indicating that no file is being used i e the conversion program called by LCMgui solely determines the starting values of the Contro
241. rom a metabolite NAA or Cr that is in the old set For example suppose that it is NAA If you do use NAA be sure to bring the pH up to 7 2 by titration with concentrated NaOH 2 Make a new Basis Set from your new Basis Spectra Give your NAA a new name e g in makebasis new in set METABO NAA2 As with all Basis Sets the new Basis Spectra must be consistently scaled with each other Suppose that the new Basis Set is called new basis Suppose your old Basis Set is called old basis Combine them with cat old basis new basis gt combined basis 3 SS 4 Run LCModel as in Sec 8 7 1 with the following extra Control Parameters BASCAL T 8 8 SUMMARY 101 CHBCAL NAA2 NCALIB 1 CHCALT 1 NAA You must also input NUNFIL DELTAT amp HZPPPM equal to those of the basis set Call the concentration output by this analysis CON Chascal 5 Modify your makebasis new in by multiplying CONC for each metabolite by CONCbascal Then run MakeBasis to produce your modified new basis 6 Repeat Steps 3 amp 4 The concentration output by LCModel should now be 1 0 7 Modify your makebasis new in by removing the complete Namelist NMEACH for NAA2 You can also simply change the NAA NMEACH e g to NM Run MakeBasis to get this final new basis without the superfluous NAA2 8 Step 3 now produces your final combined basis 8 8 Summary The main steps in processing your Model Spectra are
242. roscopy in temporal lobe epilepsy Absolute quantitation with partial volume correction for cerebrospinal fluid Proc Intl Soc Magn Reson Med 6 1726 1998 P B Barker B J Soher S J Blackband J C Chatham V P Mathews R N Bryan Quantitation of proton NMR spectra of the human brain using tissue water as an internal concentration reference NMR Biomed 6 89 94 1993 B J Soher R E Hurd N Sailasuta P B Barker Quantitation of automated single voxel proton MRS using cerebral water as an internal reference Magn Reson Med 36 335 339 1996 G Jost I Harting S Heiland Quantitative single voxel spectroscopy The reciprocity principle for receive only head coils J Magn Reson Imaging 21 66 71 2005 E R Danielsen O Henriksen Absolute quantitative proton NMR spectroscopy based on the amplitude of the local water suppression pulse Quantification of brain water and metabolites NMR Biomed 7 311 318 1994 E R Danielsen T Michaelis B D Ross Three methods of calibration in quantitative proton MR spectroscopy J Magn Reson B 106 287 291 1995 G Helms Analysis of 1 5 Tesla proton MR spectra of human brain using LCModel and an imported basis set Magn Reson Imaging 17 1211 1218 1999 G Helms A precise and user independent quantification technique for regional compar ison of single volume proton MR spectroscopy of the human brain NMR Biomed 13 398 406 2000 T Ernst R K
243. rrent correction button is active blue the File Selector window appears unless you have Marconi Picker files or GE Probe P files which already contain the water reference It is exactly as in Fig 6 1 except for the window of instructions at the upper left Fig 6 5 As you can see from Fig 6 5 the filename of your currently selected water suppressed data is shown and you must select the corresponding unsuppressed data file With Philips Siemens Toshiba and GE spectrum G files if the voxel locations of the two files do not coincide then a window appears with an error message and with options for continuing or starting over LCMegui cannot always extract accurate voxel positions from Siemens syngo files and this warning is sometimes a false alarm if you are sure that the voxels match then you can simply click the button Do Wrong Water to continue with the analysis Missing data If LCMgui is not able to extract all necessary information from the data files new windows will prompt you for these values 6 5 WHERE IS THE OUTPUT 51 Overwrite archive files LCMgui automatically archives the output in an Archive Directory that is usually unique for each data file However if you want to analyze a data file for a second time a window will appear warning you that the old output is about to be overwritten You are given the opportunity of aborting changing the output directory or overwriting it Save changes If you have chang
244. s it is not necessary for concentration ratios Chapter 11 discusses fine points and further Control Parameters that will be useful for some users Chapter 12 explains all of LCModel s diagnostic messages and gives general hints for troubleshooting Even as a lowly Normal User you may still want to at least look at the Table of Contents at the front of this manual to get an overview of all the possibilities of LCModel It is easy to modify the LCModel Control Parameters to make use of the possibilities described in Chaps 5 9 amp 11 LCMgui users may also want to learn about a Control Parameter or other term that is usually set automatically for them You can often look up such a term using the Index at the end of this manual 1 2 Normal Use of LCModel With LCMgui you generally only have to click on a few choices such as the name of the file containing the data to be analyzed Instead of using LCMgui the person installing LCModel could also write a simple Pre Formatting Program or script to automatically format the input data and then run LCModel Then the Normal User only has to input the name of the raw time domain data to be analyzed It is not even necessary to view the data before analysis phasing referencing and quantitation are done automatically by LCModel There is also another LCModel output file archiving the results which are then available for computing means stan dard deviations trends etc in the metabol
245. s whose sum is usually studied such as GPC PCh or NAA NAAG are inconsistent Then the normally much more reliable sum will have increased scatter It is therefore important to calibrate members of such pairs with each other The concentration of the Cd complex of GPC from Sigma is particularly unreliable To calibrate GPC with PCh input the following Control Parameters which are defined below BASCAL T CHBCAL GPC NCALIB 1 CHCALI 1 PCh PPMST 3 45 You must also supply the usual HZPPPM NUNFIL DELTAT FILBAS FILPS FILRAW You can supply the name of any valid RAW file for FILRAW since the data will be the GPC Basis Spectrum PPMST cuts off the resonances above 3 45 ppm which are different in GPC amp PCh Normally one singlet of each of the two metabolites is used for calibration PPMSHF 8 7 CALIBRATING BASIS SPECTRA 99 Sec 11 3 3 might be needed in cases where the CHBCAL singlet must be greatly shifted to the CHCALI singlet When BASCAL T the data Basis Spectrum is divided by its CONC Thus when the number of protons contributing to the peak is the same for CHBCAL and CHCALI 1 as here the LCModel concentration will be 1 0 if the two are consistently scaled Therefore you correct the GPC CONC in your makebasis in file by multiplying it by the LCModel CONC that comes out of this calibration analysis Check this by running MakeBasis with this corrected makebasis in and then running the
246. s HWDPHA 0 15 If any of your DSS peaks are so broad that they have not decayed into the Baseline within this interval then you should repeat your acquisitions You can check this by running with PPMST 0 15 and PPMEND 0 15 94 PPMSCA PPMPHA PPMBAS SDPNTS 8 6 4 3 CHAPTER 8 MAKING THE BASIS SET REAL The chemical shift accurate to within HWDSCA ppm of your reference peak for scaling Default PPMSCA 8 44 REAL The chemical shift accurate to within HWDPHA ppm of your reference peak for phasing Default PPMPHA 0 0 REAL ppm Baseline The Baseline in the integration of the formate peak is estimated from two intervals PPMBAS ppm wide on each side of the integration interval specified in the definition of HWDSCA above Default PPMBAS 0 1 Defined in Sec 8 5 2 Namelist NMEACH There is one of these for each Basis Spectrum to be included in the Basis Set It sets parameters for the particular Basis Spectrum being processed FILRAW METABO DEGZER DEGPPM CONC CONCSC NOSHIF CHARACTER 255 the name of the RAW file containing the ECC time domain data for the Basis Spectrum to be processed CHARACTER 6 the abbreviation of the metabolite In this manual it is called the Metabolite Name This is used throughout LCModel for iden tification You must use the Metabolite Names in the Allowed Name column of Table 8 1 If two Metabolite Names are given in Table 8
247. s starting with Removed show deletions that you have made Click Save Changes below to save these changes to a new modified Control Defaults file In later sessions you can then optionally select this or another Control Defaults file to automatically set the defaults to these values so that you do not have to enter them again Click on Do Not Save Changes to continue this analysis using your changes but not saving them Save Changes Do Not Save Changes Abort LCMgui hzpppm 63 87 sddegz 6 0 doecc T Figure 7 8 You can save all the Control Parameters including your modifications for optional use as starting defaults in the future l New Control Defaults Filename AE In the green field below enter the new Control Defaults filename where your changes are to be saved In the window at the bottom are filenames of any Control Defaults files that already exist Be careful not to enter one of these names in the green field unless you want to overwrite the old file Typical special Control Defaults and their filenames might be longTE or calibrate with NAA New ename SES Click on Save below to save your changes in the green file You can then select these Control Defaults later by clicking on Advanced Settings in the Control Parameters window Click on Use Next Time below to save your changes in the green file and to also automatically use these as defaults in the future Click on
248. se the Two Spectra Method i e with water suppression For several of your representative spectra use the Preview Data button in Fig 7 2 Sec 7 4 1 to display the absolute value spectrum between 7 ppm and 1 ppm If the residual water peak is strongly distorted e g with an extra peak or shoulder then use DOWATR F below Otherwise it is best to let LC Model fit the residual water peak since it will tend to handle its overlap with Lip53 Lip52 better than if this water peak is cut out using DOWATR F Consistently use the same value for DOWATR for an entire study DOWATR LOGICAL DOWATR F will not analyze the data between 5 0 ppm and 3 6 ppm 4 0 ppm when SPTYPE liver 11 Default DOWATR T 9 3 1 4 Measurements at room temperature ROOMT LOGICAL Set ROOMT T if your VOI is around room temperature e g ex vivo Default ROOMT F 9 3 LIPID SPECTRA LIVER BREAST BONE ETC 109 9 3 2 Output See PLOT 11 in HOME 1cmodel doc figures pdf The main output is in the top right Concentration Table All concentrations are measured as resonance areas of the metabolite The term resonance area is used instead of peak area since LCModel attempts to resolve overlapping signals even when they form only one combined peak The main differences between the four SPTYPE are which metabolites are present in the analysis Lip13 lipid signal around 1 3 ppm Lip09 lipid signal around 0 9 ppm usually difficu
249. sing or an unex pected character instead of the end of the string was encountered 3nn metwt is missing or non numeric or an unexpected character instead of the end of the string was encountered Ratio of baseline to solution If the distance traveled by the baseline is large compared to the maximum peak in the solution a more detailed search for phase corrections is made 1 FATAL A critical optimization has failed to converge Hint Check for poor quality data 2 FATAL Same as RBASOL 1 3 warning All concentrations are zero at an intermediate stage in the anal ysis This probably means that the data are very poor regula fals adjusts the overall regularization until the criterion in 11 for choosing ag and as in 1 Eq 4 is satisfied Problems here are often restricted to one of many independent analyses and do not necessarily affect the final results 1 ILLOGICAL Meaningless arguments in call to RFALST 2 FATAL At least one of the Hidden Control Parameters RALIMN or RALINC is less than 1 0 Hints Do not input these 3 FATAL A critical unregularized least squares analysis has failed Hints Check for very poor or erroneous data 4 info ag has reached its maximum allowable value Hints This is normal with very smooth Baselines or with very noisy data 5 info ag has reached its minimum allowable value Hints Working Hard 6 info ag has reached its maximum allowable value Hints This is comm
250. sis and VITRO T is set Furthermore when all CHCALI are members of the usual set Lac NAA Cr or Cre GPC PCh Cho then the Control Parameters NREFPK PPMREF amp HZREF are automatically set Then the only extra Control Pa rameters for calibration that you need to input are NCALIB amp CHCALI and if you want to calibrate a Basis Spectrum BASCAL amp CHBCAL 8 7 2 Adding New Basis Spectra This section specifies how you can add new Basis Spectra to an old Basis Set As mentioned in Sec 8 3 2 if all Basis Spectra were processed using Auto Scaling and Auto Phasing with formate and DSS then you simply run them all together with one makebasis in file They are all scaled together using their CONCSC values for the respective formate concentrations Of course the sequence parameters must al ways match including HZPPPM DELTAT amp NUNFIL The linewidth of your new spec tra must match the old one given in ppm by FWHMBA near the beginning of the BASIS file How do you determine if an old BASIS file was produced with Auto Scaling amp Auto Phasing The information is in the Namelist NMUSED Auto Scaling was used only if AUTOSC T amp AUTOPH T appear in every Namelist NMUSED You will then also find CONCSC for each Basis Spectrum in its NMUSED The rest of Sec 8 7 2 specifies how you can add new Basis Spectra to an old Basis Set that was not Auto Scaled The steps are 1 Also acquire one new Basis Spectrum f
251. some parts are clearly noted to be only of interest to users of LCMgui and many parts only to non users The only necessary reading for everyone is the Disclaimer on page 2 and Chaps 1 3 LCMgui users will also find Chapter 6 useful It is also assumed through out this manual that you are familiar with Ref 1 but not necessarily its Appendix This is sufficient for most users referred to here as Normal Users it should allow you to competently use LCModel Chapters 4 amp 5 need only be read by the person initially installing LCModel in your laboratory Chapter 5 describes important input Control Parameters specific to your spectra such as field strength dwell time etc With LCMgui these are often automatically determined Chapter 6 is a guide on the basics of using LCMgui It is highly recommended for LCMgui users Chapter 7 is a detailed reference manual on LCMgui It can help you fully exploit the flexibility of LCMgui and optimally install and configure it for your site and your Normal Users Chapter 8 need only be read by the person constructing the Basis Set of in vitro model metabolite Basis Spectra for your laboratory This is not necessary if your sequence matches one of the Basis Sets that are freely available check with me first 11 12 CHAPTER 1 PREFACE amp OVERVIEW Chapter 9 describes further useful Control Parameters Chapter 10 need only be read if you want to estimate absolute metabolite concentra tion
252. st be in column 2 followed by the Namelist name e g SEQPAR above The end is specified by END Most convenient is that the Namelist can be very long but only the variables explicitly input in the Namelist are changed the rest keep their original default values Most Control Variables have default values and these are specified in this manual If you agree with the default of a Control Parameter then you do not have to input anything for it Exceptionally none of the Control Variables in Namelist SEQPAR has a default value The above Control Parameters are ECHOT REAL echo time the echo time in ms used for this data HZPPPM REAL Hz per ppm the field strength in terms of the proton resonance frequency in MHz i e HZPPPM 42 58Bp with Bo in Tesla You should input it with an accuracy of at least four significant figures SEQ CHARACTER 5 must be either PRESS or STEAM according to the localiza tion method used for this data 5 2 2 Namelist NMID The second Namelist in the RAW file is NMID and it is always required An example is NMID ID pat77 777 FMTDAT 2E15 6 VOLUME 18 0 TRAMP 89 6 END The above Control Parameters are 36 CHAPTER 5 RUNNING LCMODEL WITHOUT LCMGUI BASIC INPUT ID FMTDAT VOLUME TRAMP BRUKER SEQACQ CHARACTER 20 a string that you can use to identify the data It ap pears in the so called Detailed Output and in the plot of the data with t
253. stic are then normal warning The first order phase correction is more than 6 standard devi ations SDDEGP from its expectation value DEGPPM The phasing may be poor and you may have to reject this analysis Hints Large artifacts or errors in the data can in rare cases cause this Much more common is that the prior information that you are imposing 12 2 STANDARD LCMODEL DIAGNOSTICS 159 12 is wrong Check for a wrong value of DEGPPM default is 0 0 or too small an SDDEGP With CSI data do not input DEGPPM or SDDEGP Let the Bayesian learn ing in LCModel set these Occasional occurrences of this diagnostic are then normal ILLOGICAL Programming error with the priors for concentration ratios FISHNI evaluates the Fisher F distribution used for determining the regularization pa rameters ag and as in 1 Eq 4 1 FATAL The number of free parameters in 1 Eq 1 is either zero or greater than the number of data points in the Analysis Window Hints Never Happened but if it does check PPMST and PPMEND for an unreasonably narrow Analysis Window FSHSSQ computes the quality of the fit sum of squared deviations to the data correspond ing to a given confidence level for choosing the regularization parameters ag and ag in 1 Eq 4 For all of the FSHSSQ diagnostics Hints Never Happened applies 1 2 3 4 5 6 FATAL A meaningless argument has been passed to FSHSSQ This could be the Hidden C
254. t effects as in Sec 8 4 Phase corrections will be applied later by MakeBasis 8 6 4 MakeBasis IN File As you can see from HOME 1cmodel test control makebasis in it consists of e Namelist SEQPAR once at the beginning e Namelist NMALL once at the beginning e For each Basis Spectrum a Namelist NMEACH 8 6 4 1 Namelist SEQPAR This is nearly the same as SEQPAR in Sec 5 2 1 they both include ECHOT amp SEQ However HZPPPM is not included here but in NMALL below In addition a new Control Parameter is included here FWHMBA REAL FWHM of Basis peaks the approximate full width at half maximum in ppm of the singlets in the Basis Set Default FWHMBA 0 013 appropriate for the test Basis Set You can initially omit FWHMBA and use your first runs of MakeBasis to determine FWHMBA by zooming in on NAA Cr GPC PCh respectively inputting PPMST PPMEND in Namelist NMALL below as 2 1 1 9 3 1 2 9 amp 3 3 3 1 You can then estimate the FWHM of these four singlets from the PS files and use the mean for FWHMBA Typical values are 0 02 0 013 0 01 for 1 5 T 2 0 T amp 4 7 T respectively A FWHMBA gt 0 025 is unacceptable it is too close to possible in vivo resolution FWHMBA is used to estimate the total FWHM for the One Page Output and the final lineshape and the smoothness of the lineshape 8 6 4 2 Namelist NMALL This sets parameters universally for all Basis Spectra in the run You must input the
255. t the end is this diagnostic with NUNFIL and its limit If you really need a higher limit then I can send another version 3 FATAL Your PPMST or PPMEND or one of the Control Parameters in INITIA 1 is so unreasonable that your Analysis Window extends outside your entire spectral window Hints Increase your bandwidth or correct your input of PPMST or PPMEND 4 FATAL Your PPMEND gt PPMST or one of the following is unrealistically small DELTAT NUNFIL HZPPPM or the number of data points in your Analysis Window exceeds the allocated storage normally 12000 Hints First check the above five Control Parameters If they are correct then I can send a version with even more allocated storage but this would only occur with unusually large fields Analysis Windows or acquisition times for one scan You probably should truncate your time data see STARTV 24 5 ILLOGICAL My computation of the machine precision shows more than 100 significant figures for DOUBLE PRECISION 6 warning There are less than 12 significant figures in your DOUBLE PRECI SION arithmetic Hints Never Happened 7 FATAL There are less than 7 significant figures in your DOUBLE PRECISION arithmetic Hints Never Happened 8 ERROR There are less than 64 points in the Analysis Window Hints Check for PPMST too close to PPMEND or too large gaps using PPMGAP 9 FATAL Gap boundaries must be strictly decreasing and not overlap i e PPMGAP J K gt PPMGAP L K 1
256. t you have done in the present Configure Save Directory window by clicking the Cancel button Your changes only take effect when you click the Set Configuration button Even after Setting Configuration you can always click on Change Con figuration to further modify or correct the configuration You can restore the original LCMegui default Save Directory configuration by clicking on Change Configuration and then on Restore Default No matter how badly you have messed things up you can always start over completely as specified in Sec 7 5 1 1 You can protect your configuration from being accidentally permanently changed as specified in Sec 7 5 1 2 7 3 3 4 Reconfiguring for File Extensions Normally the PS file is a file called ps in the Archive Directory The following Siemens example using Fig 7 5 shows how to archive them as extensions to filenames 1 2 3 Click the green Append button next to Default Save Directory Click the green Append button next to PatientName Enter _ underscore in the green field next to Enter a fixed element and then click the Append Entry button to the right Click the green Append button next to Tail of Data SetId In the green field next to Enter a fixed element delete the underscore and enter period then asterisk and then click the Append Entry button to the right This is th
257. ta file You can further check this by moving the mouse without 50 CHAPTER 6 ELEMENTARY GUIDE TO LCMGUI a Instructions Male Select your Siemens UNSUPPRESSED WATER REFERENCE file corresponding to the following water suppressed file that you just selected fhone sp tmp mni 0302 Fras_ms rda Instructions 1 Click on each sub directory and press Enter 2 Double click on your data file Figure 6 5 The instructions part of the File Selector for the unsuppressed water reference clicking it to the Reload Data button at the bottom A balloon message will appear showing the filename that you selected If this file is incorrect you can click on the Reload Data button the File Selector window will appear and you can start over selecting a new data file i e you start again at Sec 6 2 1 6 4 Run LCModel Once you are satisfied with the settings in the Control Parameters window you click the yellow Run LCModel button at the bottom Normally the analysis is immediately started as a background process and a new window appears offering you the chance to start another analysis by selecting a new data file and proceeding again from Sec 6 2 1 You can start a new analysis before previous analyses have finished as each analysis finishes its results are automatically output However four windows may appear before the analysis can start Select unsuppressed water reference file If the Do eddy cu
258. tabonomics 148 MRSI data sets 20 137 multi channel data 145 multi user installations 54 multi voxel data sets 20 37 38 49 137 muscle spectra 103 my bin2raw LCMeui 72 N1HMET 129 NAIM 89 NAMEAC 120 Namelist 35 NAMREL 119 NCALIB 100 NCOMBI 118 NDATAB 98 NDCOLS 39 182 NDROWS 39 NDSLIC 39 NEACH 120 NEXT2 149 NKEEP 117 MALL Namelist 92 MEACH Namelist 94 MID Namelist 35 MKECC Namelist 86 MUSED Namelist 98 NORAT 145 NOT2 142 NOMIT 117 non standard data files LCMgui 72 NORATO 145 Normal Users 11 NOSHIF 94 NRATIO 143 NREFPK 136 NSDSH 149 NSDT2 149 NSHIFT 137 NSIMUL 140 NSUBTK 122 NTITLE 151 NUNFIL 38 NUSE1 140 NVOXSK 139 222242242424 off resonance spectra 134 offsets Baseline 95 One Page Output 15 Other button LCMgui 27 Other data types 27 OWNER 37 PAGEHT 121 PAGEWD 121 partial volume correction 131 PDF files 30 PGNORM 41 phantoms 43 140 phase corrections GE 74 phased arrays 145 phasing conventions 88 phenylalanine 148 Philips data 25 36 phosphorus P 152 INDEX Picker data 25 41 58 62 95 PLABSV 88 PLFREQ 88 PLOT nn in figures pdf 13 Plot customizing 120 Plot One Page Output 17 120 PlotRaw diagnostics 89 PlotRaw program 87 plotting RAW files 87 PLTIME 88 PLTRAW Namelist 87 PLTRAW diagnostics 89 PostScript to PDF conversion 30 PPMAPP 95 PPMBAS 94 PP
259. te8_300mhz_070 basis steam press The appropriate string must be in the filename File Extension This must be basis BASIS or Basis 7 3 3 Archiving the LCModel Results You can specify which type of output will be archived by LCMgui and where it will be archived It is obviously best to organize this data base as soon as possible 7 3 3 1 Which File Types will be Archived Clicking on the Save File types in the Control Parameters window Fig 7 2 produces the menu in Fig 7 3 You click on a filetype to make its button blue to have LCMgui archive it in this and each future session By default the TABLE file a compact tabular summary of the results useful for later statistical and data base 7 3 BASIC SETTINGS AND USAGE 61 CI Save File types E E TABLE compact tables EPs necessary PostScript output Mcs CSV format for Spreadsheets CONTROL input Control Parameters F COORD coordinates for plots PRINT detailed output dump M CORAH corrected output RAW file gt FRAN water suppressed input M H20 unsuppressed water input Figure 7 3 You get this menu when you click the Save File Types button in Fig 7 2 analyses and the PS file with the One Page Output are archived See the Index at the end of this manual for the other filetypes in Fig 7 3 7 3 3 2 In which Directory will they be Archived The directory where these files will be saved is shown in the green f
260. th Philips data in the post proc page setting Spectral correction to no will switch off ECC 9 2 2 4 Water Scaling For Water Scaling you must also acquire an unsuppressed water reference spectrum from the same voxel immediately before or after the water suppressed spectrum LCMgui users can activate the Do water scaling button in Fig 6 4 Sec 6 3 4 Otherwise you could input DOWS T Bruker users must use Sec 10 2 2 1 Users with all other scanners must also use Sec 10 2 2 1 if the total gains in the suppressed and unsuppressed spectra are different However in the correction factor use only the Gain ratio not the NS ratio which is only needed with Bruker You can set ATTH20 to this correction factor without the 0 7 factor since the default value of ATTH20 is 1 0 with muscle 5 9 2 MUSCLE SPECTRA 105 Section 9 3 2 1 specifies how to convert from resonance areas to less convenient concentra tions 9 2 2 5 Referencing You only need to read Sec 9 2 2 5 if the LCModel output spectrum is shifted to an incorrect referencing shift LCModel uses the Cr peak for referencing If the Cr peak at 3 03 ppm is not clearly dominant in this region you could instead try referencing to the residual water peak provided it is within about 0 05 ppm of 4 65 ppm You then input DOREFS 1 T DOREFS 2 F If the water peak cannot be used and the Cho peak around 3 2 ppm is higher than the Cr peak at 3 03 ppm you c
261. the Analysis Window you simulate Glyc simply by inputting into LCModel Sec 11 7 NSIMUL 14 Under the following other conditions long TE Asp GABA Glc GPC indistinguishable from PCh amp NAAG a shifted NAA can simulate NAAG are unnecessary high fields Considerably more metabolites can be quantified above 4 7 T especially at short TE 16 The metabolites in the bottom part of Table 8 1 occur in pathologies and some may be necessary for your studies It is better to measure too many than too few metabolites You should include all metabolites that you will conceivably ever want to use With Control Parameters in Sec 9 8 1 you can easily exclude irrelevant Basis Spectra from an analysis and for consistency it is best to acquire all Basis Spectra in one session rather than adding new ones later Whenever possible use salts rather than acids in the interests of better pH control The order numbers in the table were kindly supplied by users who obtained spectra with no apparent contamination They are the latest that I have but most are outdated Reports of contaminated metabolites are listed below For example it seems important to use L Lactic acid and L Alanine D L mixtures of Lac from several sources were contaminated The Basis Spectra must have both higher accuracy and resolution than the n vivo spectra Higher concentrations are useful in achieving this The recommended con centrations are in column 3 NAAG
262. the ratio of the metabolite resonance area to the unsuppressed water resonance area The 3rd column has the heading L16 LO for ratios to the total resonance area of Lip16 Lip09 Lip13 you can change to other ratios using Sec 9 9 1 2 Without Water Scaling the Conc column has the resonance areas generally in unknown useless units If PPMST lt 5 0 the 3rd column has the heading L16 LO for ratios to Lip16 Lip09 Lip13 If PPMST gt 5 0 the 3rd column may instead 110 CHAPTER 9 FURTHER USEFUL OPTIONS AND INFORMATION have the heading Water this is only meaningful if you are using the One Spectrum Method with no water suppression Obviously PPMST around 5 0 is never used it is either at least 7 0 to fully include the water signal or not exceeding 4 0 to exclude the water signal Please also note the following e As with all peak fitting methods in contrast to LCModel analyses of brain restricted empirical models are used and the SD values are much more strongly underestimated Sec 11 1 Therefore only SD values be low 5 are output in boldface e Some analyses can take much longer because of the search for optimal phase corrections The diagnostic info TWOREG 12 signals a very de tailed search e These analyses require more work than brain and info diagnostics about Working Hard Sec 12 2 are common as in PLOT 9 9 3 2 1 Resonance areas vs concentrations Two of the three cases above give you Rati0ar
263. the screen On a Linux platform no other computationally intensive process should be running since the elapsed time is measured With Unix Sun DEC Compaq or SGI this is less critical since the LCModel CPU time is usually measured Unix computers typically take about 150 seconds Recent Linux PCs typically take about 5 seconds Chapter 5 Running LCModel without LCMgui Basic Input Although all users would benefit by becoming familiar with the Control Parameters in the chapter LCMgui users with GE Marconi Picker Philips Siemens Toshiba amp Varian data can skip this chapter LCMgui does all this for you Only LCMgui users with other data types using the LCMgui Other button Instead of using LCMgui you can run LCModel directly with a command like HOME 1cmodel bin lcmodel lt my control where my control is a CONTROL file Sec 5 3 containing changes to the Control Parameters This chapter specifies the necessary Control Parameters that you may have to set for the conditions in your laboratory It also outlines how you would write a simple Pre Formatting Program or script to automatically do this and to connect files to LCModel The Normal User then only needs to input one name identifying the data 5 1 Conventions 5 1 1 File extensions LCModel filetypes are denoted by extensions below For clarity this manual uses upper case LCMgui generally uses lower case except for RAW you do not have to follow eithe
264. the test run results into HOME 1cmodel test output test ps You can then transfer this file to a com puter that can print or display the file You could also convert the PostScript files to PDF files using Adobe Acrobat Distiller okular or ps2pdf present in some Linux distributions If you later find a display or print command that works you can build this into LCMgui using Sec 7 4 6 4 2 4 Further Tests Without a license and without the test data LCModel will stop with an error mes sage but a plot of the absolute value of the spectrum without a referencing shift correction will still be output Therefore this plot will indicate whether your data have been correctly converted and read by LCModel You can continue with your first session or you can exit and start LCMgui again with the following commands cd HOME 1cmodel 1cmgui You can test the automatic conversion of your data as follows e Click on the Further tests button when the run with the LCMgui test data is finished or click on Test My Data when you restart LCMgui later e In Fig 6 3 click on the data type corresponding to your data See Sec 3 7 e In Fig 6 1 select your data file See also Sec 6 2 1 e LCMgui should automatically convert your data and plot the absolute value of the unreferenced spectrum in the LCModel One Page Output Keep in mind that the absolute value of a spectrum has much poorer resolution than the real part
265. tion than your in vivo spectra Output the file HOME 1cmodel test output basis ps with the Basis Spectra from the test run in Chap 4 You should aim at this high quality higher quality at 9 4T and above Although LCModel can handle weaker water suppression in vivo as discussed in Sec 3 4 your water suppression n vitro must be good enough so that the formate and DSS singlets have nearly flat Baselines in order for Auto Scaling and Auto Phasing to be reliable The high concentrations in Table 8 1 can help the S N You can also do more scans but only as long as the resolution loss due to frequency drifts is certain to be negligible Similarly do not use an excessively large voxel It is important to use the same voxel size for each Basis Spectrum and to shim before every model spectrum The resolution must be better than you will ever achieve in vivo at low fields singlet linewidth FWHM less than 0 03 ppm The resolution must also be consistent FWHMs within about 0 005 ppm of each other among the Basis Spectra Check this by plotting the phased spectra or by running MakeBasis Repeat some acquisitions if necessary 8 3 2 Consistency It is essential that the Basis Spectra are scaled consistently with each other other wise even the concentration ratios will be incorrect The Auto Scaling capability of MakeBasis should greatly facilitate this If you do not use Auto Scaling then you must be especially careful that all the spect
266. tive Hints Plot the absolute value of the spectrum with PlotRaw Sec 8 5 or with LCMgui s Preview Data Sec 7 4 1 to see if there is a well defined formate peak in the spectrum Another possibility is that a big referencing error due to too weak a DSS referencing peak shifted the formate peak away from PPMSCA An enormous shift at least NUNFIL Grid Points has been computed with XTRASH Hint Plot the absolute value of the spectrum with PlotRaw Sec 8 5 or with LCMgui s Preview Data Sec 7 4 1 You should see that your values for XTRASH or perhaps PPMAPP or PPMPK are grossly in error PPMAPP is right at the edge of the spectrum or you forgot to input a positive PPMFLA Hints Increase the bandwidth or correct the input of PPMAPP or PPMFLA FLATEN amp PPMFLA are obsolete and should no longer be used You must have PTTITL amp PTLABL positive and RHTITL amp RHLABL at least 1 0 Same as ONEPAG 4 in Sec 12 2 Same as ONEPAG 8 in Sec 12 2 PPMSCA is too close to the edge of the spectral range Hint Increase the bandwidth or correct the input of PPMSCA PPMPHA is too close to the edge of the spectral range Hint Increase the bandwidth or correct the input of PPMPHA You must have PPMOFF 1 gt PPMOFF 2 and the difference must be sufficient to include at least 5 Grid Points PPMPHA is right at the edge of the spectrum Hints Increase the bandwidth or correct your input of PPMPHA Same as AXRND1 1 2 3 in Sec 8
267. tribute to this signal 29 Similarly some Cr CH3 signal may be spread to around 2 8 2 95 ppm roughly es timated by cr28 and to around 3 1 3 2 ppm where it overlaps with the signal from cholines total roughly estimated by cho IMCL CH signal around 0 9 ppm EMCL CH signal around 1 1 ppm water signal this value is useless in the usual case of water suppression In addition a residual suppressed water signal is sometimes so distorted that the models cannot fit it well IMCL signal around 5 3 ppm EMCL signal around 5 5 ppm total signals similar to IMCL3 and IMCL4 in 30 Fig 2 in the region 1 3 ppm mainly visible as a broad hump around 2 0 2 5 ppm These are important to avoid overestimating EMCL15 amp IMCL13 as simple peak integration would do total signals similar to EMCL3 and EMCL4 in 30 Fig 2 in the region 1 3 ppm mainly visible as a broad hump around 2 0 2 5 ppm These are important to avoid overestimating EMCL15 amp IMCL13 as simple peak integration would do total signal possibly Tau in the region 3 3 3 6 ppm Cr CH signal around 3 93 ppm but this is often strongly distorted and there are often other poorly defined signals in this region Please also note the following As with all peak fitting methods in contrast to LCModel analyses of brain restricted empirical models are used and the SD values are much more strongly underestimated Sec 11 1 Therefore only SD values be
268. trum ppm range The convolution range will be reduced Hints Either your spectrum has very poor resolution or your bandwidth is too small There seem to be a large number of spectral data points per peak That is the Preliminary Analysis has estimated a lineshape so wide that an Ns in 1 Eq 1 would effectively exceed NSIDMX 11 Grid Points The increment n in Sec 9 9 4 will be made 2 3 or 4 info or greater warning to cover the convolution range in no more than NSIDMX steps Hints With a warning only The following can only be used for single voxel data not CSI or multi channel data If the data quality and resolution are acceptable and you consistently get this warning for all your spectra then display your time data with LCMgui s Preview Data Sec 7 4 1 If these time data have completely decayed into the noise in the first half or quarter of the time range then truncate the remainder that is mainly noise You simply input an NUNFIL 1 2 or 1 4 the original NUNFIL there is no need to change the RAW file Bruker and Varian users can use the Preprocessor Sec 7 4 5 truncate to automatically reduce NUNFIL You must consistently do this for all your data all your spectra must have the same acquisition time for one scan warning The Preliminary Analysis has estimated an excessively broad line shape Hints Check for very poor data or spectral resolution FATAL Your spectrum does not extend far enough on the ppm ax
269. uisition time for your Basis Spectra than you normally use for your in vivo spectra The bandwidth must be high enough i e the dwell time DELTAT small enough so that the formate and DSS peaks are not near the edge of the spectrum Otherwise these can be so distorted by your scanner s anti aliasing filter that the Auto Phasing fails If the center frequency is at water the bandwidth must be at least 16 ppm At 1 5 T this corresponds to 1000 Hz DELTAT 0 001 s A bandwidth of 12 ppm causes Auto Phasing to fail you would have to phase the spectra yourself manually 8 3 4 Repetition Time Acquire all Basis Sets with TR 10000 ms The basis spectra are then fully relaxed and you only have to correct for the T relaxation in vivo You can and should avoid these T relaxation effects in vivo at a small cost in S N by using long TR in vivo at least 4000 ms 8 4 Eddy Current Correction Eddy current correction ECC is automatically done by Philips if you set spectral correction to yes on the post proc page you can then skip Sec 8 4 For each spectrum you should acquire an unsuppressed water reference from the same voxel This is automatically done for you with GE Probe P files and all Marconi Picker files The rest of Sec 8 4 specifies how you can do the ECC with or without LCMgui 8 4 EDDY CURRENT CORRECTION 85 8 4 1 ECC with LCMgui During one LCMgui session you will probably do a series of ECCs one for
270. uming that there is more than one column Scanner displays are not always consistent with this Sec 11 4 3 2 5 3 CONTROL FILE 39 You specify the CSI data set dimensions of the RAW file with the following Control Parameters NDCOLS INTEGER Number of data columns The number of columns in the CSI data set Default NDCOLS 1 NDROWS INTEGER Number of data rows The number of rows in the CSI data set Default NDROWS 1 NDSLIC INTEGER Number of data slices The number of slices in the CSI data set Default NDSLIC 1 You specify the range rectangular subset of voxels to be analyzed by LCModel with ICOLST amp ICOLEN INTEGER Voxels in columns ICOLST through ICOLEN will be analyzed Default ICOLST ICOLEN 1 IROWST amp IROWEN INTEGER Voxels in rows IROWST through IROWEN will be ana lyzed Default IROWST IROWEN 1 ISLICE INTEGER Voxels in slice ISLICE will be analyzed Default ISLICE 1 There will now be a set of output files for each voxel and these are uniquely identified as follows the Slice only for 3D sets and Row amp Col are inserted at the beginning of each TITLE as in PLOT 15 in HOME 1cmodel doc figures pdf The output filenames are extended as in the following example if you input FILPS gt filename ps then the Slice 1 Row 2 Col 8 file will be filename_s11_2 3 ps Section 11 4 3 lists other cases and also specifies how to change the format of t
271. ur RAW files When AUTOPH T you must input AUTOSC T HWDPHA must correspond to at least 2 Grid Points you must not input DEGZER or DEGPPM and PPMSCA and PPMPHA must be at least 1 0 ppm apart When AUTOPH F you must input DEGZER and DEGPPM You must have SDPNTS lt 60 You forgot to input PPMAPP or you input PPMAPP 1 lt PPMAPP 2 or a PPMAPP is so extreme that it comes too close to the edge of the ppm range of the data to also allow proper smoothing Hints If your reference peak is really near the edge increase the bandwidth Oth erwise correct your input of PPMAPP In the entire window between PPMAPP 1 and PPMAPP 2 either a the real part of the spectrum is negative where it should actually have the referencing peak Hint This might be just a negative offset that could be remedied by adding a positive number to the first time domain data point or b the absolute value of the spectrum is zero Hint Check for gross errors in your RAW file that could cause the data to be all identically zero An enormous shift at least NUNFIL Grid Points has been computed Hint Plot the absolute value of the spectrum with PlotRaw Sec 8 5 or with LCMgui s Preview Data Sec 7 4 1 you should see that your values for PPMAPP or PPMPK are grossly in error PPMSCA is right at the edge of the spectrum Increase the bandwidth or correct your input of PPMSCA The area of the real part of the formate peak at PPMSCA is nega
272. urther topics and options Scan the section headings Topics that are of more restricted interest or that have been included mainly for completeness are in smaller print 11 1 Error Estimates amp Reproducibility The standard error estimates SD Sec 2 1 are also called Cram r Rao lower bounds from the Cram r Rao inequality In practice the most important limitation of these lower bounds is not the nonlinearity of the model or the finite amount of data mentioned in the textbooks that makes them lower bounds A far bigger problem is that these lower bounds and most statistical tests assume that the model is correct or at least sufficiently parameterized to describe the data and this is seldom true in biological systems Methods that fit say a linear baseline plus a Lorentzian to each peak give too optimistic lower bounds because the models are wrong over simplified There are usually more complicated lineshapes amp baselines and each peak is generally a mixture of several metabolite signals And regardless of the quantitation method there are numerous possible systematic errors such as relaxation and partial volume effects So although these SD are useful guides all of these limitations and addi tional sources of error must be kept in mind Reproducibility can also be misleading For example the LCModel baseline can automatically account very well for a missing or incorrect macromolecule model with good
273. ut metabo ppm amp amp are used e The tokens FWHM AMP must be exactly as above with only spaces between the entries i e each CHSIMU must be in one line with no new line despite the incorrect multi line printing of CHSIMU 7 amp CHSIMU 8 above Notes e The numbers after Lip amp MM above indicate the approximate ppm value of the main peak e The most important signals of Lip20 amp MM20 are considered to be around 142 CHAPTER 11 FINE POINTS 2 ppm The respective AMP values in this region sum to 2 0 to estimate mM of CHa groups The minor peaks at 2 8 amp 3 ppm are counted separately e The easiest way to omit some of the above simulated Basis Spectra from the analysis is to use CHOMIT amp NOMIT e At short TE you should have PPMEND 0 2 to include MM as well as the lipids If PPMEND gt 0 6 most MMs and lipids are omitted This can be overridden using Sec 9 8 2 e If you acquire your own basis set do not acquire any singlet spectra simulate them For example for Scyllo input NSIMUL 14 CHSIMU 14 Scyllo 3 35 0 FWHM 9 lt 0 0 AMP 6 NNOT2 INTEGER is the number of metabolites to be omitted from the Final Analysis i e they are only used in the Preliminary Analysis Default NNOT2 3 CHNOT2 CHARACTER 300 6 the first NNOT2 elements contain the Metabolite Names of the Basis Spectra to be excluded from the Final Analysis Default CHNOT2 1 Lip13c CHNOT2 2 Lip13d
274. ve added SDDEGZ amp HZPPPM at the bottom Case is not important for the names of Control Parameters only strings within single quotes are case sensitive With GE Probe raw P files and Marconi Picker data eddy current correction ECC is automatically done by LCMgui Otherwise during your first runs you should decide whether you can do ECC Sections 9 6 amp 9 7 discuss ECC and the settings of SDDEGZ Y SDDEGP However LCMgui automatically sets the Control Parameter FILH20 mentioned in Sec 5 3 4 for you During one of your first runs you should also enter PGNORM in this window See the Index at the end of this manual for terms like PGNORM If you print then you should use a duplex printer otherwise you will also want to enter IPAGE2 0 7 3 5 Saving Your Control Parameters If you have made any changes to Control Parameters the window in Fig 7 8 will appear after you click Run LCModel Only the modified Control Parameters are shown but you are offered the chance of saving all the Control Parameters for optional use in the future You can click the Save Changes button This produces the window in Fig 7 9 In Fig 7 9 you enter a name for the file that will store your Control Parameters This is called a Control Defaults File In Fig 7 9 I have entered the name g files ecc 66 CHAPTER 7 LCMGUI REFERENCE MANUAL zj Save Changes 243 Below are listed additional changes to Control Parameters that you have made Line
275. w you can select and display a rectangular subset of a slice and have LCModel analyze this subset or the whole slice in one multi voxel run An advantage of this is that LCModel first analyzes a central voxel of the subset and then works outwards learning starting estimates and soft constraints for first order phase corrections and referencing shifts from the preceding often better central voxels for the often poorer outer voxels This can speed up and improve the analyses You should still be selective with your subset analyses of poor quality voxels can take several times longer than usual Without LCMgui the Control Parameters in Sec 5 3 2 specify the dimensions of the CSI data set and of the rectangular subset 138 CHAPTER 11 FINE POINTS 11 43 Multi Voxel Filenames You input a single TITLE and single output filenames but there will be one for each voxel These are made unique as follows The Slice only for 3D sets and Row amp Col are inserted at the beginning of each TITLE A voxel identifier is inserted into every output filename For Slice 2 Row 3 amp Column 4 the default identifier is s12_3 4 as used by Alger et al 32 The Postscript output file is then s12_3 4 ps 11 4 3 1 Customizing Output Filenames The placement of the identifier depends on the form of your input pathname For example FILPS normally ends with ps There are three allowed cases dir ps default for LCMgui dir s12_3 4 ps dir fi
276. when L gt J checks whether the curves to be plotted on the One Page Output have been phased well enough to correspond to the real part of the spectrum 1 ILLOGICAL Meaningless arguments for LDEGMX 2 FATAL One of the Hidden Control Parameters DEGMAX 1 or DEGMAX 2 is not positive Hints Do not input these 12 2 STANDARD LCMODEL DIAGNOSTICS 161 LESSMO LOADCH MAIN MAKEPS MERGLF MERGRT MYBASI Checks if the Preliminary Analysis has found concentrations of the metabolites named in CHLESS that are unexpectedly large compared to the metabolite named in CHMORE If so the Preliminary Analysis is then repeated with the CHLESS metabo lites omitted For the first three diagnostics the Hint is CHLESS amp CHMORE are Hidden Control Parameters and should not be input 1 FATAL There is an unreasonable number of metabolites named in CHLESS 2 ERROR One of the metabolite names in CHLESS equals that in CHMORE LESSMO will be skipped 3 ERROR Same as LESSMO 1 LESSMO will be skipped 4 info LESSMO will repeat the Preliminary Analysis with the CHLESS metabo lites omitted loads the Input Changes table for the One Page Output 1 ERROR You attempted to set the Hidden Control Parameter MERMES to exceed allocated storage It will be reduced to the maximum allowed value of 600 2 FATAL Error trying to re read Namelist LCMODL in your CONTROL file Hints System error in rewinding file Other errors should have alr
277. y output in this way Section 7 4 2 specifies how you can both print and simultaneously display the results on your monitor You can also completely avoid printing by displaying only and using the archived PS and TABLE files for later reference Most display programs allow you to print anyway Data File Type In the second test Sec 4 2 4 during installation you select the data type in Fig 6 3 and test the automatic conversion and input of data to LCModel If you do not have the automatically convertible file types specified in Sec 3 7 then Sec 7 4 4 specifies how you use your own data conversion script with the Other button in Fig 6 3 7 2 4 Install Model Spectra You need a Basis Set of model metabolite spectra compatible with your data as specified in Sec 8 1 We have a large collection of basis sets especially for PRESS freely available for use with LCModel Be sure to contact me first before acquiring your own using Chap 8 To install the basis sets first unzip them e g with gunzip and then be sure to mv them to the directory HOME 1cmodel basis sets 7 2 4 1 Multiple Field Strengths You can skip this unless you have data acquired at more than one field strength LCMgui does not use field strength to select default Basis Sets Here is a detailed 56 CHAPTER 7 LCMGUI REFERENCE MANUAL l Install License 2153 Congratulations on your purchase of LCModel In the yellow field enter the exact OWNER string t
278. y part of the complex pair is acquired one sample time after the real part must you input SEQACQ T in this case you input the uncorrected data as the usual complex pairs and all 4 programs in the LCModel package automatically correct the data for you So the only necessary input in this Namelist required from everyone is FMTDAT if you are only interested in concentration ratios For absolute concentrations LCModel needs a TRAMP that scales the in vivo data consistently with the Basis Spectra VOLUME then accounts for the voxel size Chapter 10 specifies this in detail In addition absolute concentrations are more sensitive than concentration ratios to the corrections for relaxation Sec 11 2 and partial volume effects which are all independent of LCModel 5 2 3 Time Domain Data These must start on the line immediately after the END of Namelist NMID These data points must be in order of increasing time with each point a complex pair real imaginary real imaginary 5 3 CONTROL FILE 37 5 2 3 1 CSI data sets The time domain data for one voxel must be immediately concatenated to the pre ceding time data with no blank positions between therefore a FMTDAT with only one complex pair per line is safest e g 2E15 6 There are also no Namelists SEQPAR or NMID for voxels after the first only the time data This is already the structure of the the time domain data in Bruker Philips Siemens Toshiba amp
279. yellow field and start constructing again So you can experiment and undo any errors by clicking Start Over If the yellow field ends with e g name then your files will be archived in name table name ps etc instead of nameftable name ps Click on Cancel to cancel your changes here and keep the old configuration Click on Set Configuration below when you are satisfied with the configuration in the yellow field This will then also be used for future default Save Directories Click on Restore Default below to restore the original default Save Directory configuration for your data type Siemens Append Default Save Directory SP sp Icmodel savedi Append Original Data Directory tmp Icm binary data siemens Enter ated temer TT Aspen Ey Append PatientName vol_2401 Appeha Tail of Datasetia 148 stat Over Constructed Save Directory PERERA MM Cancel Set Configuration Restore Default Figure 7 5 The window for permanently reconfiguring the structure of Siemens Archive Directories Advanced Settings qa Change LCMgui Settings Figure 7 6 You get this menu when you click the Advanced Settings button in Fig 7 2 7 3 BASIC SETTINGS AND USAGE 65 View Edit Control Paraneters ZA Below are the Control Parameters that will be used in the CONTROL file If necessary you can change add or delete these in the window below Click on OK
280. your basis file from 2 to the values from the basis file from 1 for FWHMBA in namelist SEQPAR all 6 values in each namelist BASIS one for each metabolite You use the resultant basis file YS 8 6 6 MakeBasis Diagnostics If the run stops before a MakeBasis diagnostic appears on your screen or a BASIS file is produced then check the suggestions in Sec 12 1 1 Normally a standard MakeBasis diagnostic of the form Aborting due to error Name nn Check User s Manual appears on your screen It can also be found at the end of the BASIS file if the BASIS file is incomplete or if some or all of the expected plots are not produced The following lists all the Name nn possibly with suggested remedies if they are not obvious MAKEBA 1 Same as PLTRAW 2 in Sec 8 5 3 2 Same as PLTRAW 3 in Sec 8 5 3 8 6 RUNNING MAKEBASIS 97 ND 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 GETPHA 1 AXRND1 1 2 3 You forgot to input DELTAT NUNFIL or HZPPPM in Namelist NMALL or you input a non positive value for one of them When you use AUTOSC T HWDSCA must correspond to at least 2 Grid Points and PPMBAS to at least 4 Grid Points You forgot to input CONC METABO or FILRAW in Namelist NMEACH or you input CONC non positive You forgot to input CONCSC in Namelist NMEACH or you input it non positive Same as MYDATA 1 in Sec 12 2 You input a non positive TRAMP or VOLUME in Namelist NMID in one of yo

LCModel & LCMgui User`s Manual - s

Contents

Download Pdf Manuals

Related Search

Related Contents

LCModel &amp; LCMgui User`s Manual - s

Contents

Download Pdf Manuals

Related Search

Related Contents

LCModel & LCMgui User`s Manual - s