CytoViva Hyperspectral User Manual: 4.8 Features
Contents
1. X Fie Edt Options Plot Function Hep i File Edt Options Plot Function Help Original Spectral Profile Filtered Spectral Profile i MA h A i A ol r y DO 700 Wavelength nm 800 goo 104 8d 900 6100 660 700 Wavelength nm Figure 4 Original Spectral Curve left Smoothed Spectral Curve right Figure 5 Original Data Cube left Smoothed Data Cube right Part 3 Savitsky Golay Filtering 1 Inthe main menu bar click File Open Image File then select AgNP 100X pixelfly 2 Inthe tool bar click the CytoViva Analysis tab click Spectral Smoothing then select the Savitsky Golay Curvefit Smoothing Figure 1 Sav Gol Curvefit Smoothing Params x File C Documents and Settings System 3 Desktoy A Spatial Dimensions Full Image 3 Number of Bands 481 Spectral Range 399 87 1000 33 nm v ENVI 4 8 SEE File Basic Tools Tei iPe Wee Classification Transform Filter Spectral Map Window Cyto iva Help gt Calibration and Correction i gt Boxcar Filter Smoothing Convert ROI to Spectral Library Width 33 shuold be odd Filter Spectral Library Savitzky Golay Curvefit Smoothing Peak Location Classifier r Degree Ea typically 1 4 Restore Previous Particle Filter Results Output Result to File Memory Figure 1 CytoViva Analysis Tab Figure 2 Sal Gov Curvefit Params box 3 Select the image in the Select input file dialog or click Open New click OK The Sa2. Peak Location Classifier Params x Peak Location nm 650 25 10 00 Smoothing Level High C Must Be Largest Peak in Spectrum Note Median Bandwidth for Input File 1 22 nm Output Rule Images Yes u Output Result to File Memory Output Rule Image Filename Figure 3 Peak Location Classifier Params Box In the Peak Location box enter the wavelength of the spectral peak of interest Next enter the Tolerance this is the range around the wavelength that you will accept for deciding to include the pixel in the classification Then enter the noise floor which should be set to a value that is below the amplitude of expected spectral peaks Next choose a Smoothing Level Choose a high level of smoothing to obtain best results Lastly You can check the box if you only want the largest peak in the spectrum to be used or leave it unchecked to accept the pixel even if there is a larger peak at a different wavelength The Median Bandwidth shown in the NOTE is the spectral resolution of the input file In the example we will accept the default values 11 Check whether you wish to save the results in memory or to a file 12 Generally select No for Output Rule Image However if you want to see how close each pixel of the image comes to having the desired peak you can select Yes If this is done two grayscale images are written to the available bands list that you can display 13 Select OK and the
3. Min Max Mean Region 1 500 600 700 800 900 1000 Wavelength Select Stat gt Filename C Program Files ITTSIDL64 1libs hooks t A ROI Region 1 Red 1750 points Basic Stats i Hean Std Band 1 oeoa Band 2 069714 Band 3 Earla Band 4 419429 Band 5 594857 Band 6 998857 Band 7 eee Band 8 901143 Band 9 teaei DJ LJ OJ LJ OJ L9 L9 Od 9 Figure 2 Spectrum from lamp recording in ROI Statistics Results 10 Now we create a Correction Spectrum from the spectrum of the lamp In the main ENVI menu select Basic Tools gt Spectral Math Click the Restore button and then click on the file correction exp This file has been installed in the folder named hook which is the first folder opened by the Restore button The expression float s1 max s1 opens in the Expressions list of the Spectral Math window see Figure 3 Click on the expression and click OK Variables to Spectra Pairings Eg Exp float 1 masfel t7 Spectral Math Previous Spectral Math Expressions floats rmax s Available Spectra list Mir Region 1 Aed 1750 points Stdev Region 1 Aed 1750 points Mean Region 1 Aed 1750 points Stdev Region 1 Aed 1750 points Max Region 1 Aed 1750 points Save Restore Clear Delete Enter an expression Add to List Map Variable to Input File Output Result to N ew Window tt UE Cancel Help Figure 3 Spectral Math input window Figure 4 Variables Assignm
4. IMPORTANT PLEASE READ CAREFULLY Cyt Viva Limited Warranty CytoViva warrants for a period of one 1 year from the date of purchase from CytoViva Inc or an authorized agent of CytoViva Inc the Warranty Period that the unmodified CytoViva adapter and or Dual Mode Fluorescence Module the Equipment when new and subject to normal use and service shall be free of defects in materials and workmanship and shall perform in accordance with the manufacturer s specifications If any component of the Equipment does not function properly during the Warranty Period due to defects in material or workmanship CytoViva will at its option either repair or replace the component without charge subject to the conditions and limitations stated herein Such repair service will include all labor as well as any necessary adjustments and or replacement parts If replacement components are used in making repairs these components may be remanufactured or may contain remanufactured materials Repair or replacement without charge is CytoViva s only obligation under this warranty This warranty is NOT transferable from the original purchaser of the Equipment Limitations Other components of the product package specifically including but are not limited to the light source s light source power transformer s and cord s liquid light guide s optical filters soectroohotometer camera s software microscope part s and motorized stage are
5. be restored by choosing Restore Previous Particle Filter Results in the drop down menu under CytoViva Particle Analysis in the main image window Saving to Table Data to ASCII causes the particle filter data to be written to a text file See Figure 5 Make sure to add a txt extension to the file so it will open correctly in NotePad or Word In both cases all of the particle data from the session is saved not just the selected particles f Filter Results Ex 1 ASCII Table txt Notepad Particle ID size DN Max wh z 146 482 000 472 3525 15 1k 374 000 474 77s 5 20 1098 00 477 220 1 1 453 000 483 339 1 1 a45 000 484 563 11 1 108 000 484 563 13 20 1122 00 487 011 6 1 104 000 488 235 4 2 110 000 507 815 12 2 110 000 524 949 10 13 510 000 531 068 1k 10 332 000 533 515 Q 1 105 000 533 515 T 1 113 000 565 334 14 13 220 000 598 377 2 5 105 000 614 286 19 1 328 000 6146 734 14 G 140 000 6 amp 0 371 5 1 169 000 769 709 Figure 5 Filter Results to ASCII File Part 1 How Selection Criteria Are Used By Particle Filter Example 1 Spectral Library Export 1 Inthe Particle Filter Review Tool you could make a Spectral Library for each particle that was found over a range of wavelengths between 475 nm and 525 nm by selecting the appropriate rows in the list Then choose Export then To Spectral Library 2 The Export particles to Spectral Library dialog appears See Figure 5 Library Export Spectral Librar
6. chose sll It contains two curves that are already in the input library but have been deemed undesirable and are to be removed Since they match curves in the input library SAM will not fail to find them In most cases where you will use the Filter Spectral Library feature the External Source library of undesired spectra are similar but not identical to the spectra you want to remove If SAM is able to match any spectra in the External Source with spectra in the Base Library the matching spectra will be removed from the Base Library After running the filter the results are shown ENVI Text Input Spectral Library C Documents and Settings System 3 Dea Number of spectra filtered out 2 Number of spectra remaining in library 2 v lt gt Figure 5 Results of the filtered spectra To view the resulting spectral library in the Available Bands List right click on file name the click Spectral Library viewer The Spectral Library Viewer box opens Click on each curve to load the spectra into a new ENVI Plot Window Below are the spectral curves from the input Base Library External Source Library and the Filtered Library see Figure 6 The Input Base Library left containing four curves two which are undesired the External Library middle containing only the undesired curves and the filtered Library right containing only the desired curves Spectral Library Plots ENVI Plot Window Spectral Profile Spectral
7. cube into absorbance or reflectance units Example 1 Stained Tissue with CNTs 1 In the tool bar click the CytoViva Analysis tab 2 Then click the Calibration and Correction tab 3 Then click the Convert to Absorbance or Reflectance see Figure 1 and the Convert to Absorbance or Reflectance box will open See Figure 2 Convert to Absorbance or Reflectance Calibration and Correction Normalize For Lamp Spectrum T Spectral Smoothing t Convert to Absorbance or Reflectance Convert to fel Abiobante C Pedlectance Convert ROT to Spectral Library Filter Spectral Library a iy Enter Output Filename Choose Peak Location Classifier firi Restore Previous Particle Filter Results Figure 1 CytoViva Analysis tab Figure 2 Convert to Absorbance or Reflectance box 4 Select the image that you want to convert using the Image Input File button from the Convert to Absorbance or Reflectance box For this example use the file Sample 100X located in the Absorbance and Reflectance Folder Click Open click OK We chose a stained sample which will show how the dye absorption spectrum can be obtained with this feature 5 For the Blank Data Source use file name Blank slf located in the Absorbance and Reflectance Folder Click Open select New file select file click open click OK 6 Check Absorbance box to select the type of conversion Note If the absorbance is choosen the Blank data source file must be a preloaded
8. warranted based on the individual original manufacturer s warranties and policies There is no warranty whatsoever on the contrast filters bulbs or the coil on the motorized stage purchased as part of the Equipment This warranty does not cover circumstances beyond CytoViva s control breakage or a malfunction that has resulted from improper Or unreasonable use or maintenance accident tampering misuse neglect improper installation modification improper maintenance or service cleaning procedures shipping or repacking of Equipment or service or parts to correct problems where such service or parts are performed or provided by anyone other than CytoViva or an authorized agent of CytoViva service required as the result of unauthorized modifications or service misuse or abuse failure to follow CytoViva s operating maintenance or failure to use items supplied by CytoViva This warranty is also void if the light source light source power transformer and cord or liquid light guide is not used in accordance with the original manufacturer s instructions recommendations or documentation Warranty service will not be provided without a dated proof of purchase Please return the Warranty Registration Card together with a copy of the original receipt within thirty 30 days of purchase It is the purchaser s responsibility to return the Equipment to the authorized agent from whom it was purchased If the Equipment was purchased directly from Cyto
9. Default RGB to Current This is a known bug which will be fixed in the future Chapter 3 Particle Filter This new feature is used to automate the selection of particles in an image The Particle Filter determines where there are particles or circularly shaped objects that match criteria given by the user The results are presented in a Review dialog box that allows for refining the particle selection criteria It is one of the most useful new features of CytoViva ENVI 4 8 1 Inthe main menu toolbar select File click Open Image File in the ENVI 4 8 folder in the particle filter file select AgNPs PF 100x Click OK Image opens in a new display 2 Inthe image menu bar select the Particle Filter Anaylsis See Figure 1 Click Particle Filter 3 The Particle Filter Params dialog box will open see Figure 2 Particle Filter Params X Background Filtering Method Sum Max 1 R Band 196 G Band 122 B Band 49 AeNP NEA File Overlay Enhance Tools uate ime See window Particle Filter Spectral Max Must Exceed 100 Restore Previous Particle Filter Results J a Valid Data Max 3000 gt Filter Particles by Size O No Yes Size Threshold pixels 20 Save Particle Data to File optional Figure 1 Particle Filter Drop Down Menu Figure 2 Particle Filter Params Box 4 Select the Background Filtering Method For this example select Max NOTE The default method Max is best to use if the par
10. Library Plots nii pn dames eN P 700 BOG 906 1000 Oo 600 700 80 906 1606 Wa y e len g th 70 FOO 800 40 1600 BOO 70 D BOD 3 Wavelength Wavelength Figure 6 Input Base Library External Source Library and Filtered library Chapter 5 Convert ROI to Spectral Library This feature expands the abilities in ENVI to create spectral libraries from regions of interest ROI associated with open images In this example we will be using the AgNP 100X Pixelfly data cube l 2 3 10 11 In the main menu bar click File Open Image File then select AgNP 100X pixelfly Expand the Zoom window and move it around the four NPs at lower right area in the main image Right click on the zoom window click ROI Tool Add circular ROIs over the three largest NPs as shown see Figure 1 The ROI tool box opens In the zoom image draw a circle around the lower left particle then double right click to complete circle and a red overlay appears In ROI tool box click New Region Draw a circle around the middle particle then double right click to complete circle and a green overlay will appear Repeat as necessary In the main tool bar click the CytoViva Analysis tab click on the Convert ROIs to Spectral Library A standard input file dialog appears Select the image file for this example choose AgNP 100X pixelfly and click OK The Convert ROIs to Spectral Library box opens and shows the ROIs that were created for this image Check a
11. Viva it should be returned postage paid along with the original dated receipt to CytoViva Inc 300 North Dean Road Suite 5 PMB 157 Auburn AL 36830 Your repaired item or replacement product will be returned to you postage paid In the event the purchaser returns Equipment to CytoViva and it is determined by CytoViva that the Equipment has been returned without cause the purchaser will be notified and the Equipment returned at the purchaser s expense DI SCLAI MER OF WARRANTIES LIMITATION OF LIABILITY THE WARRANTIES CONTAINED HEREIN ARE IN LIEU OF AND CYTOVIVA EXPRESSLY DISCLAIMS AND CUSTOMER WAIVES ALL OTHER REPRESENTATIONS AND WARRANTIES EXPRESS OR IMPLIED STATUTORY ARISING IN THE COURSE OF DEALING OR PERFORMANCE CUSTOM USAGE IN TRADE OR OTHERWISE INCLUDING WITHOUT LIMITATION IMPLIED WARRANTIES OF MERCHANTABILITY TITLE INFRINGEMENT OR FITNESS FOR A PARTICULAR PURPOSE WHERE NOT PROHIBITED BY LAW CYTOVIVA WILL NOT BE RESPONSIBLE FOR ANY CONSEQUENTIAL OR INCIDENTAL DAMAGES RESULTING FROM THE PURCHASE USE OR IMPROPER FUNCTIONING OF THIS EQUIPMENT REGARDLESS OF THE CAUSE SUCH DAMAGES FOR WHICH CYTOVIVA WILL NOT BE RESPONSIBLE INCLUDE BUT ARE NOT LIMITED TO LOSS OF REVENUE OR PROFIT DOWNTIME COSTS LOSS OF USE OF YOUR EQUIPMENT COST OF ANY SUBSTITUTE EQUIPMENT FACILITIES OR SERVICES OR CLAIMS OF THIRD PARTIES FOR SUCH DAMAGES CYTOVIVAS LIABILITY WILL NEVER EXCEED THE PURCHASE PRICE OF ANY DEFECTIVE PRODUCT OR PART Thank y
12. Viva Analysis Tab 3 Select the image in the Select input file dialog or click Open New click OK The Moving Average Filter Params box will open Figure 2 Keep all image bands to avoid the mistake of creating files of differing numbers of bands 4 Enter a value for the Smoothing Width A larger value averages the curve over a longer portion 5 Save your new data to either a file Click Choose Output Filename window opens Enter file name click open Click OK Moving Average Filter Params fx File C Documents and Settings System S3 Desktoy A Spatial Dimensions Full Image Number of Bands 481 Spectral Range 399 87 1000 33 nm lt gt v Smoothing Width 5 must be odd Output Result to File O Memory Output Filename 8 examples S moothing 4qNp 1 OO pixelfly Adfilter Figure 2 Moving Average Filter Params box 6 The filtered data cube will open see Figure 3 7 On the left is the original image and on the right is the smoothed image see Figure 4 To differentiate the images in the title of the smoothed image the heading Moving Ave has been added 8 The spectral curves of a single pixel Z axis profile are shown for the original and filtered data below see Figure 5 and ASKAN p s l D X ER Moving Ave Band 195 G Moving Ave B si O Fe Overlay Brice Took Cytaea Ratha Anahit Wande 1 Spectral Profile AgNp 100X pixelfly 2 Spectral Profile AgNp 100X pixelfly
13. cle Filter Results Figure 1 CytoViva Analysis tab Figure 2 Lamp Spectrum Button Next click the Lamp Spectrum button to enter the Correction Spectrum Use NORM lamp SL for the example click Open then click OK Keep the default Data Output Type Click Choose Give the name of the output file using the input file name with COR appended to it COR reminds you that the file is a corrected version of the data Click open Click OK A corrected data cube file is saved and is opened in the available bands list and displayed Now open the original input image in the main menu bar click File Open Image File and then select the file in this example Ag 100X It will be displayed in the Available Bands List select the file right click and select Load True Color lt to new gt NOTE The corrected image will look virtually identical with the input image Use the Square Root Enhancement for the best image In the menu bar of the input image select Tools then link then Link Displays The Link Displays dialogue box opens see Figure 3 10 Select Yes for both displays and Off for Dynamic Overlay Click OK Link Displays Display 1 Yes Link sotii Jyoti Display 2 Yes Jt Link sotti yoni Link Size Position Display 1 Dynamic Overlay Off 4t Transparency 0 1 ooo S Figure 3 Link Display Dialogue Box 11 Open a Z axis profile for both images by right clicking on the image and selecting Z Profile As you cl
14. cted data and change the relative amplitudes of the features Part 2 Recording the Lamp This step is described using the CytoViva dark field condenser 1 Prepare a calibration slide with two samples under different cover slips Put the samples close to each other Use clean glass slides a The first sample 1 should be blank with just media and coverslip on glass b The second sample 2 should include particles under the cover slip that are stable and that you normally have available c Add a small amount of immersion oil to the cover slip above each sample 2 Begin by aligning the CytoViva condenser and 10X oil objective using the particle sample sample 2 Make the adjustments just as you would for normal viewing of the sample 3 Switch to the 100X oil objective Adjust for best focus Then open the objective iris fully You will see a bright field of illumination 4 Move the stage to sample 1 blank sample You will again see the bright illumination this time with no contribution from particles The steps that you have taken insure that the glass and oil are present just as they would be for actual recordings 5 Record 50 lines from the blank sample using the normal HSI settings Make sure the camera exposure Is short enough that the plotted line is not clipped in the preview mode Normally 0 005 seconds is a good camera exposure time for recording the lamp but you may need to decrease or increase the exposure so that the max
15. ectral Library to Filter dialogue box opens The user selects a spectral library file from the input list or opens a new one from this window using the Open tab see Figure 2 This file should contain all of the spectra from the analysis including both the desired and undesired spectral curves e E gad Ingat Fikes ENVI 4 8 SEE a jal File Basic Tools Eemu eee Classification Transform Filter Spectral Map Window CytoViva Help faahi TOS pect Calibration and Correction gt Spectral Smoothing gt Convert ROI to Spectral Library Filter Spectral Library Peak Location Classifier Restore Previous Particle Filter Results Figure 1 CytoViva Analysis Tab ese e Figure 2 Spectral Library to Filter box Example 1 Simple Separation of Spectral Curves 1 This example will remove two spectra deemed undesirable from a spectral library Once the Select the Spectral Library to Filter dialogue box is open select the file sl If the input file isn t in the list click open and then select the file The input file sl contains four spectra see Figure 2 located in the ENVI 4 8 Examples folder Click OK 2 The Filter CytoViva Spectral Library box appears see Figure 3 3 Inthe Filter CytoViva Spectral Library box See Figure 3 starting at the top the input file name is shown in the Base Library box Filter CytoViva Spectral Library Spectral Data Base Library C Documents and S
16. ent window OF Queue Cancel Help Clear 11 Assign the spectral curve labeled Mean Region 1 to the expression by clicking on it in the Variables Used in Expression list See Figure 4 Change the Output Result to New Window and click OK The Correction Spectrum opens in the Spectral Math Result Window 12 Save the Correction spectrum as a Spectral Library using the File gt Save Plot As tool Use the file name Correction Spectrum SL which identifies it as the current correction spectral library Save it in your imaging folder where it can be accessed easily with the Normalized for Lamp Spectrum feature Note the maximum value of the curve is 1 0 see Figure 5 Ly Spectral Math Result BOC BOG foo aod J he Figure 5 Correction spectrum The Correction Spectrum will be used to correct data cubes so that the unique spectral properties of the lamp Camera response and spectrograph do not influence the spectrum of the sample material Before correction the relatively low strength of the illumination and low response of the camera at both ends of the wavelength range produce an uneven response in the sample spectrum This uneven response Is removed from the sample spectrum after correction 13 The file you have just created is used as an input to the ENVI 4 8 CytoViva Analysis feature Normalize for Lamp Spectrum which is described below Part 3 Convert to Absorbance or Reflectance This feature converts a data
17. ettings System S Desktop Spatial Dimensions Full Image Number of Bands 492 Spectral Range 400 12 1000 09 nm lt i Number of Output Bands 50 r Average Bandwidth 12 00 nm Output Result to Fie Memory Output Filename Figure 2 Boxcar Filtering Params box 6 A new image opens On the left is the original image and on the right is the smoothed image see Figure 3 In the original image right click click Z profile Spectral profile will open select pixel of interest Repeat for new image The spectral curves of a single pixel Z axis profile are shown for the original see Figure 4 and filterer data see Figure 5 7 1 Spectral Profile AgNp 100X pixelfly Spectral Profile ea i Bul FO Wavelength Figure 3 Spectral Profile for original Image Spectral Protile Figure 3 Original and Filtered Image Figure 4 Spectral Profile for Boxcar Filtered Image Part 2 Adjacent Band Averaging 1 Inthe main menu bar click File Open Image File then select AgNP 100X pixelfly 2 Inthe tool bar click the CytoViva Analysis tab click Spectral Smoothing then select the Adjacent Bands Averaging Figure 1 ENVI 4 8 SEE Cytoviva Analysis Calibration and Correction gt Boxcar Filter Smoothing Convert ROI to Spectral Library Filter Spectral Library Savitzky Golay Curvefit Smoothing Peak Location Classifier r Restore Previous Particle Filter Results Figure 1 Cyto
18. ettings System 3 Desktop sl External Source OROls Spectral Library Image Processing Params SAM Threshold SAM Computations Floating Point Double Precision Display Report on Completion No Yes Output Files Output Result to File Memory Output Filename for Filtered Library Figure 3 Filter CytoViva Spectral Library box 10 11 12 1 Spectral Profile AgNp 100X pixelfly ox Next choose the external source source of undesired curves You can choose from an ROI Spectral library or an Image The source for this example is Spectral Library Next is the Processing Params Change the SAM Threshold from 05 to 1 The SAM threshold can be decreased if matches with the external source need to be more precise Default settings should be used for the rest of the input The Display Report on Completion will produce a window showing how many spectra were removed from the input library and how many remain Save the new library to memory for review until the desired outcome is achieved then save it to a file Click Choose create a file name for filtered library Click Open then click OK The Select external source Spectral Library box opens Select which input file to use for the external source see Figure 4 Click OK Select external source Spectral Library File Infor ghey pt e Prev J open Figure 4 Selecting the External Source input file In this example
19. ick on different particles in the input image the cursor moves to the same location in the corrected image since these images are linked The curve in the plot windows correspond to the Same pixel in both images You will see how Lamp Normalization changes the input spectra Turn on the Collect Spectra option to show multiple curves on the same plot NOTE When the input image Is normalized by the lamp spectrum spectra in the new image are corrected for the uneven spectral response of the instrument This compensates for the relatively weak illumination and camera response at each end of the instrument s spectral range The input and the corrected spectra are most alike where the lamp spectrum is near its peak which is between 500 and 600 nm 12 For the example with silver nanoparticles See Figure 4 the input image is shown arrows point at nanoparticles with spectra of interest Figure 5 shows the uncorrected spectra from the nanoparticles on the left and corrected spectra of the same particles on the right Corrected Curves m el ot oe E a ee AG 400g ae m m m 200G i AGG i Team a i E Pe T e 000 179m d F 1000 100G GLI OJ ALI g 10 SIG GCU OJ 0 g L 104 Wavelength nm Wavelength nm Figure 5 Left uncorrected spectra of particles Right corrected spectra using lamp normalizing feature NOTE Correction can bring out the presence of new spectral features that were masked in the uncorre
20. imum intensity value on the Y axis is between 1000 and 2000 Pixelfly PCI camera or between 4000 and 8000 Pixelfly USB camera and Andor cameras 6 Name the file lamp and append today s date to the file This file is an HSI data cube containing a pure recording of the lamp If the halogen light source is used every day it is best to record the lamp once per month since aging of the lamp can affect the spectrum of the output With light to moderate usage a new lamp recording should be made every three months If your halogen light source has an aluminum reflector lamp installed the output dial should be set to 3 4 full output 3 o clock setting to minimize heating of the light guide Be sure the output of the lamp is set to the normal value used for recordings Save the calibration slide for later use 7 After the HSI files opens in ENVI you will see an image of a bright field of pixels Use the ROI tool to draw a region of interest in the center of the image see Figure 1 _ amp 9 2 R Band 194 G Band 123 B Band 50 testlamp E il gt Figure 1 Image from the lamp recording 8 Anew region appears in the ROI list Click on Stats 9 A plot of the lamp spectrum appears in the ROI Statistical Results window see Figure 2 Right click and select Plot Key The white curve labeled Mean Region is the spectrum of the lamp that you will use 7 ROI Statistics Results testlamp Select Plot gt Clear Plot
21. ion and Correction Spectral Smoothing Convert ROI to Spectral Library Filter Spectral Library Peak Location Classifier Restore Previous Particle Filter Results p Figure 1 Cytoviva Analysis Tab in ENVI 4 8 Chapter 2 Calibration and Correction Part 1 Correction for the I nstrument Spectral Response The CytoViva Analysis feature Normalize for Lamp Spectrum is used to correct data cubes for the uneven spectral response of the microscope optics This process uses a recording of light from the lamp and thus the feature is termed Normalize for Lamp Spectrum in ENVI 1 Zi 3 In the tool bar click the CytoViva Analysis tab Then click the Calibration and Correction tab Then click the Normalize for Lamp Spectrum see Figure 1 and the Normalize for Lamp Spectrum box will open see Figure 2 Select the image that you want to correct using the Input Image button in the Normalize for Lamp Spectrum box Click Open then click new file for this example select the Ag 100xX file Click OK If the file is already in the Input list select the file click OK Normalize for Lamp Spectrum Calibration and Correction Normalize For Lamp Spectrum 9 Spectral Smoothing i Convert to Absorbance or Reflectance Lamo Spectum Convert ROI to Spectral Library Filter Spectral Library Dapa Date Type 2 Finning Poet O Saged irsya C reign niega Creer Output Pleneme Chigi Peak Location Classifier Restore Previous Parti
22. lamp spectra if reflectance is chosen the blank data source must be from a perfect reflector 7 Click Choose to enter an output file name By convention use the Input file name with ABS appended to the name These appendages remind you that the data has been converted to absorbance or reflectance units 8 Click Open Click OK 9 The Input file will be converted and the new image will automatically open in the Available Bands List and in a display Below are the Zoom image from the original scan and the converted absorbance scan see Figure 3 1 Zoom 4x ox 2 Zoom 4x Sele aa ver m Figure 3 Original Image left and Converted Absorption Image right 10 The spectrum of the original sample at the cross hair and the absorption spectrum at the cross hair of the converted image see Figure 4 1 Spectral Profile Sample 100X CO 42 Spectral Profile Sample 100xABS OX a aa T vectral Profile S ectral Profile BOG BOG 70G BoG j 1606 BOG BoC Wavelength Aaa AnA Ana ARAN FOO BOG 900 1000 wavelength Figure 4 Original Image Spectra left and converted Absorption Image Spectra right Note If you try to open the original datacube after running this feature you will need to manually assign it to a new display You will see that the converted image opens in the new display You get the original image back by right clicking on the file name in the available bands list and selecting Load
23. le then select AgNP 100X pixelfly In the main tool bar go to CytoViva analysis tab select Spectral Smoothing click Boxcar Filter 3 Select Input File box opens Select from list or choose Open New Click OK see Figure 1 NO S ENVI 4 8 DER Restore Previous Particle Filter Results oo000 Calibration and Correction b BA Filter Spectral Library sac E s Band 2 401 0806 Band 5 404 7216 o Band 8 408 3639 Berd 6120077 uenitee wel ese Classification Transform Filter Spectral Map Vector Topographic Radar Window Help N AEA a P Boxcar Fiter Smoothino Jie Bande List Convert ROI to Spectral Library Adjacent Band Averaging Peak Location Classifier Ee Np OOK pixelfly Band 1 399 8673 Band 3 402 2941 Band 4 403 5078 o Band 6 405 9355 o Band 7 407 1496 Band 9 409 5784 Band 10 410 7930 O Gray Scale RGB Color Band 195 638 2467 4gNp 1 00 ooo ct Band 123 549 0733 AqNp 100 Band 49 458 2872 AgNp 100 lt pi seo OS E E Mam 0 x EAE Figure 1 CytoViva Analysis Tab The Boxcar Filtering Params box opens see Figure 2 The number you will enter determines how many bands the new data will have A larger number will cause a smaller degree of noise filtering Select number of output bands or use the default of 50 Click Choose 5 Output Filename window opens Enter file name click open Click OK Boxcar Filtering Params File C Documents and S
24. ll boxes See Figure 2 For Spectra to Include select All Pixel Spectra amp ROI means This will save spectra from every pixel in each ROI along with the mean spectrum of each ROI Accept the other default settings Save the output to a spectral library file Click choose enter filename click open Click OK Convert ROH to Spectral Library k Figure 1 Zoom window with the ROIs red green and blue Figure 2 Convert ROIs to Spectral Library The file will show up in the available bands list Right click on the file and select the Spectral Library Viewer to open the spectra into list see Figure 3 View the spectra by clicking the first entry which opens the first spectrum into a plot window The rest of the entries can be plotted by either hitting the keyboards down arrow or in the menu bar on the spectral plot by going to menu bar in the Spectral Library Plot box selecting Input Data and selecting Spectral Library see Figure 4 The Mean is the average of each ROI and the regions listed below are what make up the Mean TE e Spectral Library Plots mA FFFEFEFFFIFFFFFFFEFFFFFF EFF PEER SERTEERTEPRPEER TERT ETERS tothe on 2 eee ate SUVESEUEEVTE STEP EEE EES TEE ma E u It 800 goo 86100 600 w ae pd th Figure 4 Spectral Library Plots of the individual Figure 3 Spectral Library Viewer Window pixel spectra from the ROIs Chapter 6 Spectral Smoothing This feature is used to reduce the noi
25. nce particles have been found using the filter the spectra of edges of those particles can be easily found 8 You can enter a save file for the results if you desire Click OK to run the Particle Filter 9 10 11 12 13 14 The Particle Filter Review Tool box opens See Figure 3 and the detected particles are circled in the main image window See Figure 4 With this example using the settings given above the Particle Filter automatically found and circled several smaller less bright particles The option to Sort the list of particles by ascending wavelength was chosen in the Review Tool window by clicking on MaxWL and then choosing Ascending from the Sort button The mean and maximum spectrum from the pixels forming the first listed particle are shown in the plot Sort options make it simple to select particles according to size wavelength or brightness Try using the different Sort options 2 4 Particle Filter Review Toot C Display Particle Overlay ED Figure 3 Particle Filter Review Tool Box Figure 4 Main Image Window for Particle Filter Export Save Dats i Delete Selected Total Paricle Court 19 EE You can export particles that you select to either a Region of Interest or a Spectral Library explained below To save the data from the Review Tool Box click Save Data Saving to Particle Data creates a file that contains the last state of the Review Tool This file can
26. ou for purchasing CytoViva products Table of Contents Chapter 1 I ntroduction Part 1 New Features of ENVI 4 8 Chapter 2 Calibration and Correction Part 1 Correction for Instrument Spectral Response Part 2 Recording the Lamp Part 3 Convert to Absorbance or Reflectance Example 1 Stained Tissue with CNTs Chapter 3 Particle Filter Part 1 How Selection Criteria Are Used by Particle Filter Example 1 Spectral Library Export Example 2 Region of Interest ROI Export Chapter 4 Filter Spectral Library Part 1 Selecting a Spectral Library to Filter Example 1 Simple Separation of Spectral Curves Chapter 5 Convert ROI to Spectral Library Chapter 6 Spectral Smoothing Part 1 Boxcar Filter Smoothing Part 2 Adjacent Band Averaging Part 3 Savitsky Golay Filtering Chapter 7 Peak Location Classifier Chapter 1 I ntroduction Part 1 New ENVI 4 8 Features This manual describes how to use the CytoViva Analytical Features for Microscopy in ENVI 4 8 The new features provide several useful capabilities for analyzing the hyperspectral images including quantitative analysis spectral filtering automatic particle analysis spectral peak classification and special processes for spectral libraries 1 When analytical features are installed the ENVI 4 8 tool bar shows a tab labeled CytoViva Analysis see Figure 1 _ DAK File Basic Tools Pee eee Classification Transform Filter Spectral Window CytoViva Help Calibrat
27. pectral characteristic l ENVI 4 8 Fie Basic Tools Ke ETES Classification Transform Filter Spectral Map Window Cytoviva Help b 2 il 10 Go to the CytoViva Analysis tab Select peak Location Classifier from the drop down menu see Figure 1 Fia belo Fie C Desumeny ond Setri byten Piekiy Dire Ea 2 Sane z x 42 ONL Sate irdegea TAP GS REE kyle Calibration and Correction Citi Standard Spectral Smoothing gt Convert ROI to Spectral Library eels mieten r ar ana Filter Spectral Library Upp Lett Gomer 1 1 Saori ioe amara vendo Cooke PCO Pool y Restore Previous Particle Filter Results Pee eons DECIDE Bon Figure 1 CytoViva Analysis tab Ca eaae Ta Shree Subari 492 492 Dandi e Figure 2 Select input image Box In the Select input image dialog use Open to browse for the HSI data cube that you wish to classify and enter it If the image was already opened it will appear in the Select Input File list For this example we will use AgNPs PF 100x See Figure 2 Select the data cube Generally you will keep the default values for Spatial and Spectra Subset You may subset the spatial size of the datacube If Spectral bands are selected be sure that they include the wavelengths of the peak that you are searching for Click OK The Peak Location Classifier Params dialog opens see Figure 3 Here you will enter several parameter values that are used to decide how to process the image
28. results of the Peak Classifier open see Figure 4 S 1 R Band 196 G Band 122 B Band 49 AgNP O X 3 Distance to Target Wavelength Memory2 x E a wF oe a Te be a Oy Ne 4 i a pm 2 Peak Location Classification Memory1 BAR 4 Wavelength at Classified Peak Memory SEE d i Figure 4 AgNP Data Cube Upper Left Classification Image Lower Left Distance to Target Wavelength Upper Right Wavelength at Classified Peak Lower Right AgNP I mage the original data cube Classification I mage Shows the pixels that are within 10nm of the desired peak wavelength of 650 25nm Distance to Target Wavelength Image shows bright regions if the pixel has a peak that rises above the noise floor The brightness of the pixel will increase as the distance of the peak from the desired value increases The value of these pixels can be read using the Cursor Location Value tool The value Is the separation of the peak from the desired wavelength in nanometers Wavelength at Classified Peak I mage shows an image where every pixel containing a peak above the noise floor is displayed on a gray scale The brightness of the pixel increases as the wavelength increases This Rule Image is useful for finding other peak wavelengths and how much variability there is in the spectrum of all objects
29. se in spectra from single pixels or regions of interest Noise reduction should be used to enhance the shape of spectral curves such that small features of the curves are more apparent It can also be used before spectral mapping In order to improve results There are three smoothing filters Boxcar Filter This filter averages the spectrum across a number of bands putting the average of each segment of the original spectrum into a new band The number of bands in the new data is reduced by this process The boxcar filter will significantly reduce the size of the filtered data cube Adjacent Band Averaging The spectrum is averaged across a number of bands and written to a new band in the output data The filter moves over by one band and repeats to create a new average which is put into a second band in the output image The number of bands in the output and input images are equal and sizes of the data cubes are the same Savitski Golay Curve Fit Smoothing Noise is removed from the spectrum by fitting smooth curves over segments of the spectrum This method removes random noise to the greatest degree It is possible for smoothed curves to have a reduced amplitude compared to the original curve amplitude Default settings for this feature should be used to best retain the original amplitude NOTE smoothing can only be done on an image Not on single spectra Part 1 Boxcar Filter Smoothing 1 Inthe main menu bar click File Open Image Fi
30. ticles have simple peaks in the spectrum This choice tells the Background Filter to use the maximum value from the spectrum at each pixel of the data cube to detect particles If you select Sum it uses the sum of all spectral bands to detect particles You will notice that the default value of the threshold parameter Spectral Max Must Exceed becomes 10000 when Sum is selected If the particles have complex spectral shapes Sum may be a better choice 5 Enter the threshold value for Spectral Max Must Exceed Intensity counts This is the minimum value that the filter must come up with using either the Max or the Sum Background Filtering method in order for a pixel to be found in a particle In this example the value was set above the baseline values of spectra at 100 6 Enter the value for Valid Data Max This is the highest value allowed for the maximum spectral value of a pixel in order for that pixel to be assigned to a particle This value stays the same for both Background Filter Methods The recommended setting is a value higher than the peak value of particles you want to find Those particles having higher peak values in the data cube will be excluded For this example we used 3000 7 Choose whether to also use the particle size as a filter parameter The default setting uses this option Small values cause the filter to exclude pixels around the edges of particle In this example the size has been set very small to 20 Note that o
31. v Gol Curvefit Smoothing Params box will open Figure 2 4 The Width sets how many bands are included in each fitting operation A large width tends to keep the amplitude of the filtered data the same as that of the original data The Degree setting determines how well the smoothed data can fit curves with narrow peaks or rapidly changing signal A small value can manage the fast changes where a large value gives better overall noise reduction It is best to keep the default for Width and experiment with the Degree 5 Save your new data to either a file Click Choose Output Filename window opens Enter file name click open Click OK 6 The filtered data cube will open On the left is the original image and on the right is the smoothed image see Figure 3 7 The spectral curves of a single pixel Z axis profile are shown for the original left and filtered right data see Figure 4 e gt 1 Spectral Profile AgNp 100X pixelfly OX Fie Edt Options Plot_Function Help Spectral Profile 2 Spectral Profile Memory5 Spectral Profile a Figure 3 Original and Filtered Image Figure 5 Smoothed Spectral Profile Chapter 7 Peak Location Classifier This features finds all pixels in an image that have a specified peak wavelength It operates like other classifier tools such as SAM The Peak Location Classifier can be used to search an image for objects or regions that have a peak in the spectrum that matches a known s
32. y Filtered Ag Particles Wavelength 400 121002 to 1000 088989 Output Result to File Memory Mean Particle 6 Output SLI Filename Mean Particle 4 Mean Particle 12 le FilterSExport Spectral Library Filtered Ag Particles Figure 5 Export Particles Figure 6 Library Viewer to Spectral Library Figure 7 Spectral Library Plots 3 In this example only the mean spectrum from each particle is going to be saved to a Spectral Library Click Choose for an output SLI file name Use the defaults for other settings Click OK The file is saved and added to the top of the Available Bands List 4 From the Available Bands List right click the file and select the Spectral Library Viewer and the Spectral Library Viewer dialogue box opens see Figure 6 The spectral curves of the particles that have wavelengths between 475 nm and 525 nm are listed Click on the Means of each particle and a spectral library plot of these is shown see Figure 7 Right click on the plot and select plot key if needed Example 2 Region of Interest ROI Export 1 Inthe Particle Filter Review Tool you could make a Spectral Library for each particle that was found over a range of wavelengths between 475 nm and 525 nm by selecting the appropriate rows in the list Then choose Export then To Region of Interest and the ROI Tools Dialogue Box opens see Figure 1 2 The ROI Tool opens with the particles listed as individual ROIS which you can appl
33. y to the image You can create different colors for the ROIs by right clicking Color and choosing Assign default colors see Figure 1 Window O Image Scroll OZoom Off ROI Name Color Pixels Red 20 1 20 A Green 18 1 18 B Particle 1 Blue 17 Wil Particle 11 Yellow 1 17 1 Particle 13 Cyan 20 1 20 Particle 6 Magenta 1 11 wv aT gt Figure 7 ROI Tool Box Chapter 4 Filter Spectral Library The Filter Spectral Library provides an automated process for removing spectra from a library that are different from the spectra that you want to match to an input file A library of spectra that have similar but not necessarily the same properties as the undesired spectra is needed The process removes the undesired spectra from the input library by performing a statistical comparison of spectral properties using the Spectral Angle Mapper SAM This is a versatile filter An example of use of the Spectral Library Filter would be to reduce a spectral library that was created from objects that have either one or both of two different attributes such as a range of particle size or peak wavelength to a library that contains objects with only one of the attributes or contains only the particles that have both attributes Part 1 Selecting a Spectral Library to Filter 1 Starting from the main menu tool bar click the CytoViva Analysis tab 2 Then click the Filter Spectral Library see Figure 1 3 The Select the Sp
Download Pdf Manuals
Related Search