Home

USER'S GUIDE

Contents

1. Figure 5 4 Sample run 1 results saved in an ASCII file i Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG y version 2 0 Department of Physics University of Kuopio FINLAND 1 Appendix A Frequently asked questions Based on the feedback and user experiences obtained from the users of the previous version of the HRV analysis software 32 we have collected here a bunch of frequently asked ques tions An answer to each question is given below the question Some of the questions are concerned with the optimal value of some analysis parameter Often these parameter values are however more or less case specific for example the length of the selected HRV data may change the preferred settings for FFT spectrum calculation Thus some of the answers might be more or less vague but hopefully still helpful When several RR interval samples have been selected what does the merge samples do When the Merge samples option is selected for the Sample analysis type the RR interval samples selected for analysis are simply merged into one sample by concatenating the samples m How to select the value of A in the smoothness priors based detrending A is the regularization parameter in the smoothness priors based detrending approach see Section 2 3 1 for details The value of this parameter changes the smoothness of the estimated trend i e a bigger value corresponds to a smoother trend As discussed in Section 2 3 1 the s
2. Sometimes the RR interval time series includes a disturbing low frequency baseline trend component Detrending options can be used to remove this kind of trend components Detrending options include removal of the first second or third order linear trend or the trend can be removed using a method called smoothness priors which was presented in 44 In the smoothness priors method the smoothness of the removed trend can be adjusted by editing the Lambda value The smoothness priors method is basically a time varying high pass filter and its cut off frequency can be adjusted with the Lambda parameter the bigger the value of Lambda the smoother is the removed trend The estimated cut off frequency for the given Lambda value is presented next to the Lambda value edit box In addition the trend to be removed from the RRI series is shown over the selected part of the RR series as a red line 4 2 2 Data browser The data browser segment shown in Fig 4 4 displays the measured RR interval series The RR interval data can be scrolled with the slider and the range of the axis can be changed by editing the Range value The RR series display mode can be changed with the button on the right hand side of the RR axis In addition the Y limit of the axis can be manually changed by editing the edit boxes on the left hand side of the axis In addition to the visualization of the RR interval data the RR interval axis displays also the artifact corrections descri
3. DE lmin Ny 3 4 Summary of HRV parameters The presented time domain frequency domain and nonlinear measures of HRV calculated by the software are summarized in Table 3 1 For each measure preferred units and a short description is given In addition a reference to the equation in which the specific measure is defined is given when possible and related references are given for some of the measures Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 4 Summary of HRV parameters 24 Time Domain Frequency Domain Nonlinear Table 3 1 Summary of the HRV measures calculated by the software Measure RR STD RR SDNN HR STD HR RMSSD NN50 pNN50 HRV index TINN triangular Peak frequency Absolute power Relative power Normalized power LF HF SD1 SD2 ApEn SampEn Da DFA Q1 az RPA Lmean Lmax REC DET ShanEn Kubios HRV version 2 0 Units ms ms 1 min u ms beats beats Description References The mean of RR intervals Standard deviation of RR intervals Eq 3 1 The mean heart rate Standard deviation of intantaneous heart rate values Square root of the mean squared differences between successive RR intervals Eq 3 3 Number of successive RR interval pairs that differ more than 50 ms NN50 divided by the total number of RR intervals Eq 3 4 The integral of the RR inter
4. SampEn Detrended fuctuations DFA a1 Detrended fuotuations DFA o2 Correlation dimension D2 23 May 2008 14 20 12 Mika Tarvainen Department of Physics University of Kvopio suunto_data sdf 11 06 2007 00 09 00 Page 2 2 Results for single samples sample 2 2 00 1153 05 06 09 095 1 105 11 55 a 65 70 RR 5 HR Resta AR Spectrum AR mocel order 16 mot actorized LF 0 04 0 15 HZ HF 015 04 H2 Total LIKE 13 Poincare Plot 06 08 1 12 14 16 18 RR 8 Jog 9 beats Results are Calculated from me non Getrendes selected RR series Kubios HRV Analysis version 2 0 Department of Physics University of Kuopio Finland Figure 5 3 Sample run 1 results for the standing period of the orthostatic test Kubios HRV y version 2 0 Y Biosignal Analysis and Medical Imaging Group BSAMIG Department of Physics University of Kuopio FINLAND 5 1 Sample run 4 sample_run_1_results t lt t WordPad Bile Edit View Insert Format Help 0860 SRA 2 Bo Y RR Interval Samples Selected for Analysis Sample 1 Sample limits s gt 24 324 Sample Analysis Type Single samples RESULTS FOR SINGLE SAMPLES GENERAL RESULTS SAMPLE 1 SAMPLE 2 Time Domain Results Statistical parameters Mean RR ms 1221 8694 STD RR ms 46 8336 Mean HR 1 min 49 2438 STD HR 1 min 2 4456 RMSSD ms 60 2684 NNSO count 107 PNNSO 43 8525 SDANN ms SDNN index ms Geometric p
5. lt Start h min s Length h min s 00 05 00 Remove trend components Method smoothn priors Lambda 500 0 035Hz Analysis Options Frequency bands VLF Hz o 004 LF z 004 015 F z 015 04 Interpolation of RR series A 00 05 40 00 03 20 00 05 00 Sample Analysis Type single samples v 00 06 Time h min s 00 10 09 00 14 40 4 Range s 737 je VIEW RESULTS ESPA Spectrum Y limits Common gt Sample 1 22 AR spectrum estimation results Interpolation rate Hz 4 Fix to g 00 PSD 57Hz AR spectrum c i 0 04 05 0 o2z 03 AR model order Frequency Hz Frequency Hz Power Power Power Power l ims 317 Figure 5 1 Sample run 1 of We select to remove the trend with the smoothness priors based method Once the detrending method is selected red lines appear over the RR interval data indicating the removed trend components The smoothness of the removed trend in the smoothness priors method can be adjusted by changing the Lambda value The smoothness priors detrending method can be compared to a high pass filter in which the cutoff frequency is determined from the lambda value bigger lambda corresponds to lower cutoff The estimated cutoff frequency of the detreding method is also shown next to the Lambda value Since we are now interested in LF and HF frequencies we wish to make sure that the detrending does not remove thos
6. of the segment spectra is calculated The selection of the window width and overlap in this method is simply a trade off between the frequency resolution and variance of the spectrum estimate The frequency resolution of FFT spectrum is roughly the reciprocal of the sample length i e the frequency resolution of the FFT spectrum of a 100 second sample is 0 1 Hz The variance of the FFT spectrum estimate on the other hand does not depend on the sample length but can be decreased by averaging several shorter samples which leads to decreased frequency resolution m How to select the AR model order for the AR spectrum estimate The AR method is a parametric method which can be used also for spectrum estimation In this method the RR interval series is modeled with an autore gressive model of specific order The roots of the AR polynomial which are actually complex conjugate pairs correspond to the spectral peaks in the AR spectrum Thus the order of the AR model has to be at least twice the number of expected spectral peaks in the spectrum In practice the order is however always higher than this minimum and the few extra roots do not disturb the spectrum estimate Even though an exaggerated model order can induce spuri ous peaks into the AR spectrum estimate and distort the results The AR model order naturally depends on the interpolation rate of the RR interval series but in many cases the default order of 16 is reasonable Why
7. 18 r giii 3 18 where the distance function d uj us is now defined as d u ur 3 19 Next an average of the term C r is taken 1 N m 1 C r Ce 3 20 On E To 3 20 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 3 Nonlinear methods 21 1 5 D log C r EDk log r Figure 3 3 Approximation of the correlation dimension D from the log r log C r plot which is the so called correlation integral The correlation dimension Da is defined as the limit value Da m lim lim eee 3 21 r 0 N logr In practice this limit value is approximated by the slope of the regression curve logr logC r 18 The slope is calculated from the linear part of the log log plot see Fig 3 3 The slope of the regression curves tend to saturate on the finite value of Da when m is increased In the software a value of m 10 was selected for the embedding 3 3 6 Recurrence plot analysis Yet another approach included in the software for analyzing the complexity of the time series is the so called recurrence plot RP analysis In this approach vectors uj RRj RRj 7 RRjs m vyr J 1 2 N m 1 r 3 22 where m is the embedding dimension and 7 the embedding lag The vectors uj then represent the RR interval time series as a trajectory in m dimensional space A recurrence plot is a symmet
8. 2b BRE SRE Se RR EG RO 3 3 1 Poimeare plot s srs s ep 4am ee Pe eh Ae a E ke eS 3 3 2 Approximateentiopy e e a aaa oes Dae E ae 3 3 3 Sample entropy s s e ra t 4866880 be ee ean aa be ee es 3 3 4 Detrended fluctuation analysis o o 3 3 0 Correlation dimensi n s s cosas oras AA AR 3 36 Recurrence plot analysis sana ee a a GD ee ae 3 4 Summary of HRV parameters lt ed ac tea easdem hiep pa da Software description 4 1 Inp t data formats sos se 2244 aa S a Oe Be e Ae a 42 The User interace secu au eur mera ae E SE ae Wee a Se ee E 4 2 1 RR interval series options ao sooo a AD Data DIOWSEE 205 4 moss rd aaa ERE ed a pa E e 42 3 Analysis options i s s e ne Ga eke ae ee ee eg Se we wee ws A ADA Results vie 2 460 o 4 a4 5 Oe ee ER RR ESS Se ee eH 10 11 13 13 4 2 5 Menus and toolbar buttons 00000 eee eee 453 Saving the results s s ei sa op eae hebhe ee ed di hiaal eed ead 43 1 ASCII 5 6 24422 et edd aw RRR REM RR Ee a Aoa IReport sheet ella 4h ORR She eye eee Oe wee 135 MATLAB MAT ile 2 choi auntie eae eee EER EDGES EES 4 4 Setting up the preferences 2 0 020 eee ee es Sample runs Sel Sample r n sa a aiat doh kb eee pe wl ee ee Rae a SO a a A Frequently asked questions Troubleshooting B l Windows specifiG o sesos e base fa ae Pe e Se ea ae a ee eS B2 Min S PEGG s saaa Gee ok Bou BOS a He ee he eek ee we A References 46 48 48 49
9. LF band have been thought to be of both sympathetic and parasympathetic origin 3 even though some researchers have suggested them to be mainly of sympathetic origin 25 The fluctuations below 0 04 Hz on the other hand have not been studied as much as the higher frequencies These frequencies are commonly divided into very low frequency VLF 0 003 0 04 Hz and ultra low frequency ULF 0 0 003 Hz bands but in case of short term recordings the ULF band is generally omitted 45 These lowest frequency rhythms are characteristic for HRV signals and have been related to e g humoral factors such as the thermoregulatory processes and renin angiotensin system 3 Even though HRV has been studied extensively during the last decades within which numerous research articles have been published the practical use of HRV have reached general consensus only in two clinical applications 45 That is it can be used as a predictor of risk after myocardial infarction 24 19 and as an early warning sign of diabetic neuropathy 5 33 In addition HRV has been found to correlate with e g age mental and physical stress and attention see e g the review in 3 The term HRV refers in general to changes in heart beat interval which is a reciprocal of the heart rate This is also the case here The starting point for HRV analysis is the ECG recording from which the HRV time series can be extracted In the formulation of the HRV 2 1 Heart beat period
10. Print Results Edit Preferences Close and Quit commands The Open Save Results Edit Preferences and Close commands work exactly as the corresponding toolbar buttons The difference between the Save and Save As commands is that when the results have already been saved the Save command automatically overwrites these results whereas the Save As command asks the user for a new file name The Quit command of the File menu is for exiting from the software The View menu includes Report sheet command which works similarly as the corresponding toolbar button Finally the Help menu includes the About HRV Analysis Software command which opens the same about dialog as the corresponding toolbar button Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 3 Saving the results 33 4 3 Saving the results The analysis results can be saved by selecting Save Results or Save Results As from the File menu or by pressing the save button on the toolbar This will open a file save dialog in which the saving type can be selected There are three different types in which the results can be saved That is the results can be written in an ASCII text file for further inspection the report sheets generated from the results can be saved in a PDF file or the results can be saved in a MATLAB MAT file 4 3 1 ASCII file When the ASCII text file is selected for the saving type the numer
11. Zbilut Dynamical assessment of physiological systems and states using recurrence plot strategies J Appl Physiol 76 965 973 1994 E J M Weber C M Molenaar and M W van der Molen A nonstationarity test for the spectral analysis of physiological time series with an application to respiratory sinus arrhythmia Psychophysiol 29 1 55 65 January 1992 J P Zbilut N Thomasson and C L Webber Recurrence quantification analysis as a tool for the nonlinear exploration of nonstationary cardiac signals Med Eng Phys 24 53 60 2002 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND
12. and QRS detection 10 Vasomotor sympathetic A CR RN HEART AV Sympa thetic Sympa HR thetic Tea Figure 2 1 The four baroreflex pathways redrawn from 41 Variation in venous volume AV left ventricular contractility VC sympathetic and parasympathetic vagal control of heart rate HR stroke volume V cardiac output CO total peripheral resistance TPR and arterial blood pressure BPa time series a fundamental issue is the determination of heart beat period 2 1 Heart beat period and QRS detection The aim in HRV analysis is to examine the sinus rhythm modulated by the autonomic ner vous system Therefore one should technically detect the occurrence times of the SA node action potentials This is however practically impossible and thus the fiducial points for the heart beat is usually determined from the ECG recording The nearest observable activ ity in the ECG compared to SA node firing is the P wave resulting from atrial depolarization see Fig 2 2 and thus the heart beat period is generally defined as the time difference between two successive P waves The signal to noise ratio of the P wave is however clearly lower than that of the strong QRS complex which results primarily from ventricular depo larization Therefore the heart beat period is commonly evaluated as the time difference between the easily detectable QRS complexes A typical QRS
13. applied in the future sessions Also the preference directories path from where the data file is searched for and in which the results are saved are preserved in memory The last nine opened data files will also appear in the File menu of the user interface and can be reopened from there All the preferences and preserved options used by Kubios HRV are saved in user specific folders Windows XP C Documents and Settings lt username gt Application Data KubiosHRV Windows Vista C Users lt username gt AppData Roaming KubiosHRV Linux home lt username gt kubioshrv where lt username gt is the name of your user profile The folder will include three files hrv_pref dat user_pref dat and HRVprefs mat The hrv_pref dat file includes all the preferences for the analysis options user_pref dat includes the user information prefer ences and HRVprefs mat all the preferences related to the usability of the software These files are created when the software is started for the first time and they will be updated whenever the preference values are edited The original settings of the preferences can be restored by deleting these files 1Note that the Application Data folder in Windows XP and AppData folder in Windows Vista are hidden by default and are not visible in the Windows File Explorer if the Show hidden files and folders is not selected from the Folder Options section of the File Explorer Kubios HRV Biosignal Analys
14. before analysis The detrending is usually based on first order 23 31 or higher order polynomial 39 31 models In addition this software includes an advanced detrending procedure originally presented in 44 This approach is based on smoothness priors regularization 2 3 1 Smoothness priors based detrending approach Let z R denote the RR interval time series which can be considered to consist of two components Z Zstat Ztrend 2 1 where Zstat is the nearly stationary RR interval series of interest Ztrena is the low frequency aperiodic trend component and N is the number of RR intervals Suppose that the trend component can be modeled with a linear observation model as trend H0 e 2 2 where H R is the observation matrix 0 R are the regression parameters and e is the observation error The task is then to estimate the parameters by some fitting procedure so that Ztrena H 6 can be used as the estimate of the trend The properties of the estimate depend strongly on the properties of the basis vectors columns of the matrix H in the fitting A widely used method for the solution of the estimate is the least squares method However a more general approach for the estimation of is used here That is the so called regularized least squares solution 6 arg min z HOP d Da H9 7 2 3 where A is the regularization parameter and Dg indicates the discrete approximation of the d th derivative
15. min s Range s 3600 88 O RARA O tare _ HF Hz 015 04 us p04 Spectrum Y limits FFT spectrum estimation results Interpolation of RR series PA Common pot Interpolation rate Hz Spectrum estimation oP FFT spectrum Window width s Window overlap AR spectrum AR model order PSD 8 Hz i PSD 5 Hz b Figure 4 3 Artifact correction a the artifact corrected series is visualized on top of the raw RR interval series b Corrected RR interval series Gis Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG Vay version 2 0 Department of Physics University of Kuopio FINLAND 1 4 2 The user interface 29 00 10 00 00 11 40 Range s 737 2 Figure 4 4 The data browser segment of the user interface Analysis Options Frequency bands wre o om LF Hz 045 H z 015 04 Interpolation of RR series Interpolation rate Hz Spectrum estimation FFT spectrum Window width s Window overlap 9 AR spectrum AR model order Use factorization Figure 4 5 The analysis options segment of the user interface samples are selected a sample can be removed by right clicking it with the mouse 4 2 3 Analysis options The analysis options segment shown in Fig 4 5 includes three subcategories Frequency bands Interpolation of RR series and Spectrum estimation All of these
16. of ApEn more regular signal The ApEn is computed as follows First a set of length m vectors uz is formed uj RR RRj 1 ioe RRj4m 1 j 1 2 N m41 3 7 where m is called the embedding dimension and N is the number of measured RR intervals The distance between these vectors is defined as the maximum absolute difference between the corresponding elements i e d uj tx max RRjin RRk4n n 0 m 1 3 8 Next for each uj the relative number of vectors uz for which d uj ug lt r is calculated This index is denoted with Ci r and can be written in the form Cre N m 1 Vk 3 9 Due to the normalization the value of C7 r is always smaller or equal to 1 Note that the value is however at least 1 N m 1 since u is also included in the count Then take the natural logarithm of each C7 r and average over j to yield 1 N m 1 j 1 Finally the approximate entropy is obtained as ApEn m r N 6 r 6 1 r 3 11 Thus the value of the estimate ApEn depends on three parameters the length m of the vectors uj the tolerance value r and the data length N In this software the value of m is selected to be m 2 The length N of the data also affects ApEn When N is increased the ApEn approaches its asymptotic value The tolerance r has a strong effect on ApEn and it should be selected as a fraction of the standard deviation of the data SDNN This selection enables the comparison of diff
17. operator This is clearly a modification of the ordinary least squares solution to the direction in which the side norm D H0 gets smaller In this way prior information Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 2 3 Preprocessing of HRV time series 14 Magnitude 0 5 Magnitude 0 0 1 0 2 0 3 0 4 0 5 Relative frequency b Figure 2 4 a Time varying frequency response of L N 1 50 and A 10 Only the first half of the frequency response is presented since the other half is identical b Frequency responses obtained from the middle row of cf bold lines for A 1 2 4 10 20 100 and 500 The corresponding cut off frequencies are 0 213 0 145 0 101 0 063 0 045 0 021 and 0 010 times the sampling frequency about the predicted trend H can be implemented to the estimation The solution of 2 3 can be written in the form 0 HTH X HTDI DH 1H z 2 4 and the estimate for the trend which is to be removed as Strend HOy 2 5 The selection of the observation matrix H can be implemented according to some known properties of the data z For example a generic set of Gaussian shaped functions or sigmoids can be used Here however the trivial choice of identity matrix H I RN is used In this case the regularization part of 2 3 can be understood to draw the solution towards the nu
18. options are con cerned with frequency domain analysis The very low frequency VLF low frequency LF and high frequency HF bands of HRV frequency domain analysis can be adjusted by editing the VLF LF and HF values The default values for the bands are VLF 0 0 04 Hz LF 0 04 0 15 Hz and HF 0 15 0 4 Hz according to 45 The default values for the bands can be restored by pressing the Defaults button A Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG Gy version 2 0 Department of Physics University of Kuopio FINLAND 4 2 The user interface 30 The RR interval time series is an irregularly time sampled series as discussed in Section 2 2 and thus spectrum estimation methods can not be applied directly In this software this problem is solved by using interpolation methods for converting the RR series into equidis tantly sampled form As the interpolation method a piecewise cubic spline interpolation is used The sampling rate of the interpolation can be adjusted by editing the Interpolation rate value By default a 4 Hz interpolation is used The spectrum for the selected RR interval sample is calculated both with Welch s peri odogram method FFT spectrum and with an autoregressive modeling based method AR spectrum In the Welch s periodogram method the used window width and window overlap can be adjusted by editing the corresponding value The default value for window width is 256 seconds and the defau
19. pNN50 x 100 3 4 3 2 Frequency domain methods 16 In addition to the above statistical measures there are some geometric measures that are calculated from the RR interval histogram The HRV triangular index is obtained as the integral of the histogram i e total number of RR intervals divided by the height of the histogram which depends on the selected bin width In order to obtain comparable results a bin width of 1 128 seconds is recommended 45 Another geometric measure is the TINN which is the baseline width of the RR histogram evaluated through triangular interpolation see 45 for details 3 2 Frequency domain methods In the frequency domain methods a power spectrum density PSD estimate is calculated for the RR interval series The regular PSD estimators implicitly assume equidistant sampling and thus the RR interval series is converted to equidistantly sampled series by interpolation methods prior to PSD estimation In the software a cubic spline interpolation method is used In HRV analysis the PSD estimation is generally carried out using either FFT based methods or parametric AR modeling based methods For details on these methods see e g 27 The advantage of FFT based methods is the simplicity of implementation while the AR spectrum yields improved resolution especially for short samples Another property of AR spectrum that has made it popular in HRV analysis is that it can be factorized into separate spectral co
20. report sheet windows for view The report sheet window includes 7 toolbar buttons and File and Page menus on the upper left hand corners of the windows The toolbar button icons and their actions are given below Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 3 Saving the results 34 Report Page 1 Bile Edit BS NALA B HRV Analysis Results RR Interval Time Series DER suunto_data sdf 11 98 2007 09 00 90 Page 1 1 Results for a single sample Time Domain Results Variable Mean AR STD RR SONN Mean Ha STD HR RMSSO INNSO pNNSO ah Ud 12 3 50 55 A o rt RR 5 HR Destsimin Frequency Domain Results FFT spectrum Welews periodogram 256 s window with 50 overtap AR Spectrum AR model order 16 not tectorized Frequency VEF 0 008 Hz LF 004 0 15 Hz Frequency VEF 500 Ha EF 0 04 0 15 Hz HF 015 04 Hz Total FIRE Poincare Plot 12 13 06 08 1 12 14 16 18 Jog n beats RESURS are calculated trom the mon cetrencded selected RR seres 23 May 2008 14 13 04 Mika Tarvainen Department of Physics University of Kuopio Kubios HRV Analysis version 2 0 Department of Physics University of Kuopio Finland Figure 4 9 The first report sheet including all the time domain frequency domain and nonlinear analysis results calculated by the software Se Kubios HRV Biosig
21. samples to be analyzed First add a new sample to the RR interval axes because we want to analyze both the lying and standing periods To do this you can simply right click the RR axes press Yes to the Add sample popup window and OK to verify the sample properties Now you will have two samples shown as yellow patches in the RR interval axes Then change the sample ranges to cover the periods or interest as shown in Fig 5 1 The easies way to change the samples ranges is to edit them with the mouse as described in Section 4 2 2 but the ranges can also be changed by editing the Start and Length values in RR interval series options segment Then check that the Sample analysis type option under the RR axis is set for Single samples Then analysis results are calculated for both samples separately If on the other hand Merge samples is selected then the two samples are first merged into one sample and the analysis results are calculated for this merged sample Since we are now only interested in the changes in LF and HF bands we wish to remove the lowest frequency trend components from the RR series These trend components affect on the time and frequency domain variables and thus by removing the trend from the data enables these variables to better describe the LF and HF variability which we are interested 40 5 1 Sample run Kubios HRV Analysis V Matlab HRV_v2_matlab data suunto_data sdf 41 Samples for analysis 2 Samplet lt
22. small value and vice versa Typically in DFA the correlations are divided into short term and long term fluctuations In the software the short term fluctuations are characterized by the slope a obtained from the log n log F n graph within range 4 lt n lt 16 Correspondingly the slope az obtained from the range 16 lt n lt 64 characterizes long term fluctuations see Fig 3 2 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 3 Nonlinear methods 1 2 log F n 1 4 1 6t 1 8 0 6 08 1 12 14 16 1 8 log n 20 Figure 3 2 Detrended fluctuation analysis A double log plot of the index F n as a func tion of segment length n a and ag are the short term and long term fluctuation slopes respectively 3 3 5 Correlation dimension Another method for measuring the complexity or strangeness of the time series is the corre lation dimension which was proposed in 14 The correlation dimension is expected to give information on the minimum number of dynamic variables needed to model the underlying system and it can be obtained as follows Similarly as in the calculation of approximate and sample entropies form length m vectors u uj RR RRj41 RRj m 1 j 1 2 N m 41 3 17 and calculate the number of vectors uz for which d u ug lt r that is nbr of uz d uj uz lt r CPG I yk 3
23. 0 00 00 11 40 1 00 i 00 06 40 08 Oo Analysis Options Time h min s Range s 737 gt Frequency bands Bl Cr om VLF Hz ears Interpolation of RR series common 4 Y Interpolation rate Hz Fix to PSD 57Hz Window width s Window overlap AR spectrum AR model order Power ims 166 642 Figure 4 1 The graphical user interface of the developed HRV analysis software used to correct artifacts from a corrupted RR interval series The user can select between very low low medium strong and very strong correction levels In addition a custom level in seconds can be set The corrections to be made on the RR series are displayed on the RR interval axis To make the corrections press the Apply button A piecewise cubic spline interpolation method is used in the corrections You can reverse the correction by pressing the Undo button or by selecting none as the correction level It should be noted that artifact correction generates missing or corrupted values into the RR series by interpolation and can cause distortion into the analysis results An example of artifact correction can be seen in Fig 4 3 In this case the analyzed RR interval sample includes two clear artifacts In order to remove these artifacts a medium level correction was selected The effect of correction can be verified from the user interface i e the corrected series is displaid on top of the
24. 45 In practice the NN and RR intervals appear to be the same and thus the term RR is preferred here The time series constructed from all available RR intervals is clearly not equidistantly sampled but has to be presented as a function of time i e as values t RR This fact has to be taken into account before frequency domain analysis In general three different approaches have been used to get around this issue 45 The simplest approach that have been adopted in e g 2 is to assume equidistant sampling and calculate the spectrum directly from the RR interval tachogram RR intervals as a function of beat number see is Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG Ney version 2 0 Department of Physics University of Kuopio FINLAND 2 2 Derivation of HRV time series 12 Derived RR intervals RR RR RR RR RR lt gt gt e gt q gt e gt ty t t t t t RR interval tachogram RR RR RR RR RR 1 2 3 4 5 RR interval series with two possible ways of interpolation RR we Ne RR RR RR e RR 27 ie PGE cane t t t t t Figure 2 3 Derivation of two HRV signals from ECG the interval tachogram middle panel and interpolated RR interval series bottom panel the left panel of Fig 2 3 This assumption can however cause distortion into the spectrum 28 This distortion becomes substantial when the variability is large in comparison with the mean level Furthermo
25. 49 Chapter 1 Overview Kubios HRV is an advanced tool for studying the variability of heart beat intervals Due to its wide variety of different analysis options and the easy to use interface the software is suitable for researchers and clinicians with varying premises The software is mainly designed for the analysis of normal human HRV but should also be usable e g for animal research Kubios HRV includes all the commonly used time and frequency domain variables of HRV The frequency domain variables are calculated for both nonparametric Fourier trans form based and parametric autoregressive modeling based spectrum estimates In addi tion several nonlinear HRV variables are calculated such as Poincar plot recurrence plot analysis detrended fluctuation analysis approximate and sample entropies and correlation dimension The Kubios HRV heart rate variability HRV analysis software is developed by the Biosignal Analysis and Medical Imaging Group BSAMIG at the Department of Physics University of Kuopio Kuopio Finland The software is a considerable upgrade to the previous version version 1 1 of the software described in 32 Changes from the previous version include significantly improved usability support for new data formats and many new or updated features Kubios HRV has been developed using MATLAB Release 2008a The MathWorks Inc and was compiled to a deployable standalone application with the MATLAB Compil
26. 623 648 1997 T Bragge M P Tarvainen P O Ranta aho and P A Karjalainen High resolution QRS fiducial point corrections in sparsely sampled ECG recordings Physiol Meas 26 5 743 751 2005 H J Braune and U Geisenorfer Measurement of heart rate variations influencing fac tors normal values and diagnostic impact on diabetic autonomic neuropathy Diabetes Res Clin Practice 29 179 187 1995 M Brennan M Palaniswami and P Kamen Do existing measures of Poincar plot geometry reflect nonlinear features of heart rate variability IEEE Trans Biomed Eng 48 11 1342 1347 November 2001 S Carrasco M J Caitan R Gonzalez and O Yanez Correlation among Poincar plot indexes and time and frequency domain measures of heart rate variability J Med Eng Technol 25 6 240 248 November December 2001 H Dabire D Mestivier J Jarnet M E Safar and N Phong Chau Quantification of sympathetic and parasympathetic tones by nonlinear indexes in normotensive rats amj 44 H1290 H1297 1998 I Daskalov and I Christov Improvement of resolution in measurement of electrocar diogram RR intervals by interpolation Med Eng Phys 19 4 375 379 June 1997 R W DeBoer J M Karemaker and J Strackee Comparing spectra of a series of point events particularly for heart rate variability data IEEE Trans Biomed Eng 31 4 384 387 April 1984 R W DeBoer J M Karemaker and J Strackee Spectrum of a series of point events gen
27. 82 492 August 1991 J Malmivuo and R Plonsey Bioelectromagnetism Principles and Applications of Bioelectric and Biomagnetic Fields Oxford University Press Web Edition 1995 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND References 52 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 S L Marple Digital Spectral Analysis Prentice Hall International 1987 J Mateo and P Laguna Improved heart rate variability signal analysis from the beat occurrence times according to the IPFM model IEEE Trans Biomed Eng 47 8 985 996 August 2000 J Mateo and P Laguna Analysis of heart rate variability in the presence of ectopic beats using the heart timing signal IEEE Trans Biomed Eng 50 3 334 343 March 2003 M Merri D C Farden J G Mottley and E L Titlebaum Sampling frequency of the electrocardiogram for spectral analysis of the heart rate variability IEEE Trans Biomed Eng 37 1 99 106 January 1990 I P Mitov A method for assessment and processing of biomedical signals containing trend and periodic components Med Eng Phys 20 9 660 668 November December 1998 J P Niskanen M P Tarvainen P O Ranta aho and P A Karjalainen Software for advanced HRV analysis Comput Meth Programs Biomed 76 1 73 81 2004 M Pagani Heart rate variability and autonomic diabetic neuro
28. In addition to the actual analysis options there are various other editable options which have mainly influence on the usability of the software Such options are e g the Range jh Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG A version 2 0 Department of Physics University of Kuopio FINLAND 12 4 4 Setting up the preferences Preferences Advanced settings User information Spectrum Estimation Opti Analysis options Interpolation of RR series Points in frequency domain FFT spectrum using Welch s Window width Advanced settings Report settings Window overlap AR spectrum AR model order Use spectral factorization Figure 4 12 Set up preferences window of the software Advanced settings category Preferences User information Paper Size Analysis options las 210 x 297 mm Advanced settings ASCIL File Settings Field delimiter Semicolon Report settings Decimal symbol Dot Custom print command C Use custom print command PA al Figure 4 13 Set up preferences window of the software Report settings category N Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG Ay version 2 0 Department of Physics University of Kuopio FINLAND 38 4 4 Setting up the preferences 39 and Y limit values of the data axis and various visualization options The values of these options are preserved in memory and any changes made to them will be
29. Kubios HRV version 2 0 http kubios uku fi USER S GUIDE October 16 2008 Mika P Tarvainen Ph D software bsamig uku fi and Juha Pekka Niskanen M Sc Biosignal Analysis and Medical Imaging Group BSAMIG http bsamig uku fi Department of Physics University of Kuopio Kuopio FINLAND a gil oo le MATLAB Copyright 1984 2008 The MathWorks Inc MATLAB is a registered trademark of The MathWorks Inc Contents Overview 1 1 System requirements 23 54 55 0 oa 8 bee ke de ee a LTL Windows gt ads i e i a a a a a ha A We WU 4 a Rok soadh Ree E 12 Installation AA wee LA hae Oo eRe da Ee BO 12 1 Windows e ico nk be ha a ee we hee Ae k eee DME graces OG e aon as BIG a Ane mare MS ar aa 1 3 Uninstallation s s e sose eo eee eae eee a eee eS Sil Windows e cs 2 2 dantededadee6b RRR eee dd a eas 132 MEMOS rra OS ee ae oe we i a 14 Software homepage si comerse E ee a a Se a 1 5 Structure of this pude ios s se ot hea a bho Hea RA de de ws Heart rate variability 2 1 Heart beat period and QRS detection o 2 2 Derivation of HRV time series a s s r s gura s auai a e o 2 3 Preprocessing of HRV time series ooun a 2 3 1 Smoothness priors based detrending approach Analysis methods 3 1 Time domain methods lt o s s s sa sred wonna med d eee ep bebe a 3 2 Frequency domain methods e 3 9 Nonlinear methods e lt s s ssa 2 ba
30. Linux specific 49 Use custom print command option in the Kubios HRV preferences See http pages cs wisc edu ghost gsview gsprint htm for more information m I have Matlab R2008a with Compiler toolbox installed on the same computer as Kubios HRV and printing does not work The entry for Matlab in the system path is usually before the Matlab Compiler Runtime entry created by Kubios HRV installer As a conse quence Kubios HRV uses the compiler runtime from the Matlab R2008a Com piler toolbox installation instead of the compiler runtime installed by Ku bios HRV installer The printing does not work in this situation because Kubios HRV cannot locate gscript mexw32 A workaround is to copy the file lt matlab_root gt toolbox matlab graphics private gscript mexw32 to lt matlab_root gt toolbox matlab graphics folder B 2 Linux specific m I receive the following error when trying to run Kubios HRV from com mand line error while loading shared libraries libmwmclmcrrt so cannot open shared object file No such file or directory This error is produced when the kubioshrv executable cannot locate MATLAB Compiler Runtime Do not use the kubioshrv executable directly to run Kubios HRV version 2 0 Instead use the start_kubioshrv sh script which sets up the MATLAB Compiler Runtime environment and then starts Kubios HRV When I try to print report sheets I get an error message in the terminal that starts with java lang N
31. T T T T 1 Time ms 0 100 200 300 400 500 600 700 Figure 2 2 Electrophysiology of the heart redrawn from 26 The different waveforms for each of the specialized cells found in the heart are shown The latency shown approximates that normally found in the healthy heart sampling frequency of the ECG is less than 500 Hz the errors in R wave occurrence times can cause critical distortion to HRV analysis results especially to spectrum estimates 30 The distortion of the spectrum is even bigger if the overall variability in heart rate is small 38 The estimation accuracy can however be improved by interpolating the QRS complex e g by using a cubic spline interpolation 9 or some model based approach 4 It should be however noted that when the SA node impulses are of interest there is an unavoidable estimation error of approximately 3 ms due to fluctuations in the AV nodal conduction time 41 2 2 Derivation of HRV time series After the QRS complex occurrence times have been estimated the HRV time series can be derived The inter beat intervals or RR intervals are obtained as differences between suc cessive R wave occurrence times That is the mth RR interval is obtained as the difference between the R wave occurrence times RR tn tn 1 In some context normal to normal NN may also be used when referring to these intervals indicating strictly intervals between successive QRS complexes resulting from SA node depolarization
32. ally into the edit box below the selection button The selected power axis options apply also for the report sheet The results for both spectra are displayed in tables below the corresponding spectrum axes Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 2 The user interface 31 VIEW RESULTS ta Time Domain Results Variable E Value Statistical Measures Mean RR 1222 1 STD RR SDNN ms 47 2 Mean HR 1 min 49 24 STD HR 1 min 2 46 RMSSD ms 60 4 NN5O count 107 pNNSO 43 9 Geometric Measures 4 11 2 59 HRV triangular index 10 208 HR beats min TINN ms 225 0 i Calculated from the non detrended selected RR series Set fixed axes limits RR 3 HR 1 min Spectrum Y limits AR spectrum estimation results Common v Ficto 3 00 PSD 5 Hz PSD sH2 o o 0 01 i 0 Frequency Hz Frequency Hz 0 1 0 2 0 3 04 Frequency Peak Power Power Power Frequency Peak Power Power Power Band Hz met n u Band Hz ms n u VLF 0 0391 166 98 VLF 0 0391 317 15 1 LF 0 0547 642 37 8 41 9 LF 0 0564 805 38 3 45 1 HF 0 2813 888 52 4 58 1 HF 0 2852 980 46 6 54 9 LF HF 0 7 LF HF 0 8 Figure 4 7 The results view segment of the user interface frequency domain results view selected The nonlinear results view shown in Fig 4 8 displays all the calculated nonlinear vari ables in one table All the variables are ca
33. anging for the previous or the next report sheet page and for changing the sheet by its page number However the Page menu is not shown if only one report sheet window is open 4 3 3 MATLAB MAT file In addition to saving the numeric results into an ASCII text file or saving the report sheet s in a PDF file the analysis results can also be saved in a MATLAB MAT file compatible with MATLAB R12 or higher The MAT file includes a single structured array variable named Res The Res variable includes the numeric results as well as the RR interval data and all the analysis options This saving option is aimed for MATLAB users and makes the further analysis or processing of the HRV data in MATLAB much easier The Res structure includes four fields which are shortly described as follows f_name File name of the analyzed data file f_path Full path for the analyzed data file CNT Basic information of the data file the field name refers to Neu roscan CNT file for historical reasons HRV Used analysis options RR interval data and all analysis results The HRV field is clearly the most essential one of these fields The HRV field includes six fields the contents of which are shortly described as follows Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 4 Setting up the preferences 36 Preferences E 2 Preferences A EE User information User informa
34. arameters RR tri index 10 208333 9 818182 TINN ms 225 0000 190 0000 926 2515 38 8103 65 0716 3 4238 25 7229 12 3 7152 Frequency Domain Results FFI spectrum AR spectrum spectrum AR spectrum Peak frequencies VLF Hz 0 039063 0 039063 0 039063 0 039063 LF Hz 0 054688 0 066406 0 089844 0 078125 HF Hz 0 281250 0 285156 0 152344 0 152344 Absolute powers VLE ms 2 164 9878 298 1608 390 9904 179 2697 LF ms 2 633 8926 747 8336 1491 4383 1111 4596 HF ms 2 883 8326 1025 3441 314 5418 230 0966 Relative powers VLF 3 9 8049 14 3946 17 7968 11 7877 LF 37 6709 36 1039 67 8861 73 0826 HF 52 5243 49 5015 14 3171 15 1297 Normalized powers LF n u 41 7660 42 1748 82 5833 82 8485 HF n u 58 2340 57 8252 17 4167 17 1515 LF HF ratio 0 7172 0 7293 4 7416 4 8304 EDR Hz 0 283027 0 217069 Nonlinear Results Poincare plot SD1 ms 43 013247 18 594375 SD2 ms 79 722321 86 249913 Recurrence plot analysis RPA Mean line length beats 8 1272 11 7661 Max line length beats 43 212 Recurrence rate REC 23 9838 34 5578 Determinism DET 96 6163 99 1659 Shannon entropy 2 8487 3 2655 Detrended fluctuation analysis DFA alpha 1 0 9246 1 3209 alpha 2 0 8887 0 8120 Others E 0 9868 1 0339 Approximate entropy ApEn Sample entropy SampEn 1 7363 1 2416 Correlation dimension D2 3 6877 3 3227
35. bed in Section 4 2 1 The selected sample s yellow patches in the RR axis can be modified with mouse as follows Each sample can be moved by grabbing it from the middle with the left mouse button and resized by grabbing it from the left or right edge You can also add a new sample to a specific location in the RR series by right clicking the RR axis The new sample will start from the clicked time instant and the length of the new sample is by default same as the previous sample After right clicking the RR axis a small popup window opens in which the sample start time and length can be accepted modified When more than one Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 2 The user interface 28 Kubios HRV Analysis V Matlab HRV_y2_matlab data test_measurement sdf BEE Artifact correction Level fr fium M Undo Samples for analysis 1 Sample 1 Start hmi Length h min s Method smoothn priors 1 00 01 40 00 10 01 Range s 3600 gt Window width s Window overlap AR spectrum AR model order PSD 8 Hz PSD 5 H2 Kubios HRV Analysis V Matlab HRV_v2_matlab data test_measurement sdf Ee Level medium v Samples for analysis 1 Remove Sample 4 Start h min s 23 21 53 Length h min s 00 20 00 1 x 23 45 00 23 83 21 00 01 40 00 10 01 Time h
36. can be done by selecting Save Results from the File menu or just by pressing the save button on the toolbar Then select Save all txt mat pdf as the save type and enter a file name You do not need to give any extension to the file name The numeric results of the analysis will be saved in the txt text file and in the mat MATLAB file and the report sheets in the pdf file The generated PDF file will now include two pages one for the results of the first RR interval sample the lying period and one for the second sample standing period These report sheet pages are shown in Figs 5 2 and 5 3 In the text file the results for the two samples are presented side by side as can be seen from Fig 5 4 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 5 1 Sample run Report Page 1 43 DER BSARLLA HO HRV Analysis Results RR Interval Time Series 23 May 2008 14 20 12 Mika Tarvainen Time Domain Results Variable Mean RA STD AR SONN Mean HR STD HR RMSSO NNSO pNNSO RA triangular index TINN Frequency Domain Results suunto_data sdf 11 08 2007 09 00 90 Page 1 2 Results for single samples sample 1 2 a 55 HR Destemin FFT spectrum Weins periodogram 256 s window with 50 overtap AR Spectrum AR model order 16 nat tectorized o B Psp mA Frequency Sano VEF 0 008 Hz LF 004 0 15 H
37. date analysis results options which have already been described in Sections 4 2 1 4 2 3 and 4 2 4 The Advanced settings category shown in Fig 4 12 includes Spectrum estimation options The spectrum estimation options include one additional option compared to those described in Section 4 2 3 i e points in frequency domain option The point in frequency domain is given as points Hz and corresponds by default to the window width of the FFT spectrum If spectrum interpolation is desired the points in frequency domain can be in creased The Report settings category shown in Fig 4 13 includes the following options The paper size of the report sheet can be changed between A4 210x297 mm and Letter 8 5x11 inch size The default paper size is A4 In addition the field delimiter and decimal point used when saving the results in an ASCII file can be selected The Custom Print Command option allows the use of an external program to print the report sheets in PostScript format All modifications for the preferences are saved by pressing the OK button Note that the OK button saves the preferences but they will be applied only in the next session A session is considered to be ended when the program is restarted or Close file is selected If on the other hand a new file is opened without first closing the previous file preferences will not be applied but the local settings changes made in the user interface are applied for the new file as well
38. detector consists of a preprocessing part followed by a decision rule Several different QRS detectors have been proposed within last decades 46 34 35 17 12 For an easy to read review of these methods see 1 The preprocessing of the ECG usually includes at least bandpass filtering to reduce power line noise baseline wander muscle noise and other interference components The passband can be set to approximately 5 30 Hz which covers most of the frequency content of QRS complex 34 In addition preprocessing can include differentiation and or squaring of the samples After preprocessing the decision rules are applied to determine whether or not a QRS complex has occurred The decision rule usually includes an amplitude threshold which is adjusted adaptively as the detection progresses In addition the average heart beat period is often used in the decision The fiducial point is generally selected to be the R wave and the corresponding time instants are given as the output of the detector The accuracy of the R wave occurrence time estimates is often required to be 1 2 ms and thus the sampling frequency of the ECG should be at least 500 1000 Hz 45 If the Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 2 2 Derivation of HRV time series 11 Common bundle Bundle branches gt Purkinje fibers Ventricular muscle T T
39. do the power values of Kubios HRV 2 0 spectrum estimates differ from those of the version 1 1 The differences in power values are due to two changes First of all in version 2 0 the mean of the RR interval data is removed before spectrum estimation This decreases significantly the VLF power value of the FFT based spectrum estimate Secondly the scaling of the spectrum estimates is changed as follows In version 1 1 the spectrum estimates were scaled such that the total power from 0 fs Hz was equal to the variance of the RR data where fs is the sampling frequency of the RR data i e the interpolation rate In version 2 0 on the other hand the total power from 0 f 2 Hz is equal to the variance Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND Appendix B Troubleshooting The Kubios HRV seems to take ages to start The MATLAB Compiler 4 has changed dramatically from the earlier versions It is now more of a deployment tool than a compiler because it only generates a wrapper executable that starts the MATLAB Compiler Runtime MCR and runs the heavily crypted Matlab M files of the compiled application on top of the MCR Although this has many advantages the main disadvantage is that the starting of the MATLAB Compiler Runtime takes roughly the same time as starting MATLAB This can be anything from 10 to 50 seconds depending on the speed of the c
40. e rhythm of the heart is controlled by the sinoatrial SA node which is modulated by both the sympathetic and parasympathetic branches of the autonomic nervous system Sympa thetic activity tends to increase heart rate HR and its response is slow few seconds 3 Parasympathetic activity on the other hand tends to decrease heart rate HR and mediates faster 0 2 0 6 seconds 3 In addition to central control there are some feedback mechanisms that can provide quick reflexes One such mechanism is the arterial baroreflex This reflex is based on baroreceptors which are located on the walls of some large vessels and can sense the stretching of vessel walls caused by pressure increase Both sympathetic and parasympathetic activity are influenced by baroreceptor stimulation trough a specific baroreflex arc Fig 2 1 The continuous modulation of the sympathetic and parasympathetic innervations results in variations in heart rate The most conspicuous periodic component of HRV is the so called respiratory sinus arrhythmia RSA which is considered to range from 0 15 to 0 4 Hz 3 In addition to the physiological influence of breathing on HRV this high frequency HF component is generally believed to be of parasympathetic origin Another widely studied component of HRV is the low frequency LF component usually ranging from 0 04 to 0 15 Hz including the component referred to as the 10 second rhythm or the Mayer wave 3 The rhythms within the
41. e consuming for longer samples and in that case it might be useful to disable the automatic update by unchecking the Automatic check box in the bottom left corner of the user interface When unchecked one or more changes to options can be made without updating breaks and when finished with changes the Apply button can be pressed to update the results The time domain results view shown in Fig 4 6 displays the time domain variables in a table and the RR interval and heart rate histograms in the two axes Most of the results are calculated from the detrended RR series if detrending is applied but there are two obvious exceptions i e mean RR interval and mean HR which are marker with the x symbol The frequency domain results view shown in Fig 4 7 displays the results for both FFT and AR spectrum estimation methods Both methods are applied to the detrended RR series The spectra of the two methods are presented in the two axes FFT spectrum on the left and AR spectrum on the right The frequency axes of the spectra are fixed to range from O Hz to the upper limit of HF band plus 0 1 Hz Thus for the default frequency band settings the frequency axis range is 0 0 5 Hz The power axes of the spectra on the other hand can be adjusted with the options on the upper left corner of the frequency domain results view The power axes can be selected to have either common or separate upper Y limits If common Y limit is selected it can also be entered manu
42. e frequencies This can be easily done by changing the Lambda value in such a way that the cutoff frequency is below 0 04 Hz The effect of detrending can also be verified by inspecting how it changes the FFT spectrum Here we set the Lambda value to 500 The time domain frequency domain and nonlinear analysis results for the selected sam ples can then be viewed in the results view segment Just make sure that the results have been updated check that the Automatic is checked in Apply changes and if not press the Apply button Press the Time domain Frequency domain or Nonlinear button to view the corresponding results Note that the results are shown only for one of the samples at a time To take a look at the results of the other sample press the lt or gt gt button on the top right corner of the results view segment the text on the left changes to indicate which Na Kubios HRV Y version 2 0 Y Biosignal Analysis and Medical Imaging Group BSAMIG Department of Physics University of Kuopio FINLAND 5 1 Sample run 42 sample s results are shown this sample will also be highlighted in the RR series axis Note that you can force a common Y limit for the spectra of both samples by setting a common Y limit value manually in the frequency domain results view For example we have here fixed the Y limit value to 0 04 s Hz Once we are done with the analysis we wish to save the analysis results in all possible formats This
43. er 4 8 The MATLAB Compiler Runtime MCR version 7 8 is required for running Kubios HRV and is included in the Kubios HRV installers lKuopio University has only limited rights to the software These limited rights are governed by a certain license agreement between Kuopio University and The MathWorks Inc 1 1 System requirements 5 1 1 System requirements The system requirements given below should be considered as recommended system re quirements The software may work also with lower system specifications but will probably function slower or with reduced usability 1 1 1 Windows e Operating system Microsoft Windows 2000 XP or Vista e Processor Intel Pentium 4 and above or equivalent 32 bit x86 processor e Memory 512 MB of RAM 1024 MB or higher recommended e Disk space about 500 MB e Desktop resolution of 1024x768 or higher e The MATLAB Compiler Runtime 7 8 included in the Kubios HRV installer 1 1 2 Linux e Operating system a Linux distribution with kernel 2 4 x or 2 6 x and glibc glibc6 2 3 4 and above Processor Intel Pentium 4 and above or equivalent 32 bit x86 processor e Memory 512 MB of RAM 1024 MB or higher recommended e Desktop resolution of 1024x768 or higher e Disk space about 460 MB The MATLAB Compiler Runtime 7 8 included in the Kubios HRV installer 1 2 Installation 1 2 1 Windows Make sure that you have administrator privileges and run the Kubios HRV installer file Follo
44. erated by the integral pulse frequency modulation model Med Biol Eng Comput 23 138 142 March 1985 50 References 51 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 G M Friesen T C Jannett M A Jadallah S L Yates S R Quint and H T Nagle A comparison of the noise sensitivity of nine QRS detection algorithms EEE Trans Biomed Eng 37 1 85 98 January 1990 Y Fusheng H Bo and T Qingyu Approximate entropy and its application in biosignal analysis In M Akay editor Nonlinear Biomedical Signal Processing Dynamic Analysis and Modeling volume II chapter 3 pages 72 91 IEEE Press New York 2001 P Grassberger and I Procaccia Characterization of strange attractors Phys Rev Lett 50 346 349 1983 P Grossman Breathing rhythms of the heart in a world of no steady state a comment on Weber Molenaar and van der Molen Psychophysiol 29 1 66 72 January 1992 S Guzzetti M G Signorini C Cogliati S Mezzetti A Porta S Cerutti and A Malliani Non linear dynamics and chaotic indices in heart rate variability of normal subjects and heart transplanted patients Cardiovascular Research 31 441 446 1996 P S Hamilton and W J Tompkins Quantitative investigation of QRS detection rules using the MIT BIH arrhythmia database IEEE Trans Biomed Eng 33 12 1157 1165 December 1986 B Henry N Lovell and F Camacho Nonlinear dynamics time
45. erent data types A common selection for r is r 0 2SDNN which is also used in this software 3 3 3 Sample entropy Sample entropy SampEn is similar to ApEn but there are two important differences in its calculation 40 20 For ApEn in the calculation of the number of vectors uz for which d uj ux lt r also the vector uj itself is included This ensures that C7 r is always larger than 0 and the logarithm can be applied but at the same time it makes ApEn to be biased In sample entropy the self comparison of uj is eliminated by calculating C7 r as 7 nbrof ug d uj up lt r Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 3 Nonlinear methods 19 Now the value of C7 r will be between 0 and 1 Next the values of C r are averaged to yield 1 N m 1 e ay 2 Cra 3 13 and the sample entropy is obtained as SampEn m r N In C r C r 3 14 The values selected for the embedding dimension m and for the tolerance parameter r in the software are the same as those for the approximate entropy calculation Both ApEn and SampEn are estimates for the negative natural logarithm of the conditional probability that a data of length N having repeated itself within a tolerance r for m points will also repeat itself for m 1 points SampEn was designed to reduce the bias of ApEn and has a closer agreement with the theory for data with
46. ic results of the analysis will be written in an ASCII text file The resulting text file includes the following information in the enumerated order Software user and data file informations Used analysis parameters Samples selected for analysis Time domain results Frequency domain results Nonlinear results Neo fF wN RR interval data and spectrum estimates The columns of the file are separated with semicolons so that the results could easily be imported to e g spreadsheet programs such as the Microsoft Excel for further inspection 4 3 2 Report sheet The software generates a printable report sheet which present all the analysis results The report sheet shown in Fig 4 9 includes all the time domain frequency domain and nonlin ear analysis results The RR interval data and the sample selected for analysis are presented on the two axes on top of both sheet and the analysis results below them If multiple analysis samples have been selected a report sheet is generated for each sample When Save Results have been selected the report sheet s can be saved in a single PDF file by selecting Report figure as the saving type in the save dialog In this case the report sheet s will not be displayed but just saved in the selected PDF file Tf you wish to view the report sheet s and or to export it into some other file format choose Report sheet from the View menu or just press the corresponding toolbar button This will open the
47. ics of Heart and Circulation pages 101 120 Institute of Physics Publishing Bristol 1993 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND References 53 42 43 46 47 48 49 50 M P Tarvainen S D Georgiadis P O Ranta aho and P A Karjalainen Time varying analysis of heart rate variability signals with Kalman smoother algorithm Physiol Meas 27 3 225 239 2006 M P Tarvainen J K Hiltunen P O Ranta aho and P A Karjalainen Estimation of nonstationary EEG with Kalman smoother approach an application to event related synchronization ERS IEEE Trans Biomed Eng 51 3 516 524 March 2004 M P Tarvainen P O Ranta aho and P A Karjalainen An advanced detrending method with application to HRV analysis IEEE Trans Biomed Eng 49 2 172 175 February 2002 Task force of the European society of cardiology and the North American society of pacing and electrophysiology Heart rate variability standards of measurement phys iological interpretation and clinical use Circulation 93 5 1043 1065 March 1996 N V Thakor J G Webster and W J Tompkins Optimal QRS detector Med Biol Eng Comput 21 343 350 May 1983 L L Trulla A Giuliani J P Zbilut and C L Webber Jr Recurrence quantification analysis of the logistic equation with transients Phys Lett A 223 4 255 260 1996 C L Webber Jr and J P
48. is and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND Chapter 5 Sample runs In this chapter we present a sample run of the software The sample run is made for the sample data file distributed with this software The sample data is measured from a healthy young male during an orthostatic test The change in the posture is known to be reflected in the low frequency and high frequency HRV in an opposite way That is when subject stands up after lying for few minutes a strong decrease in the HF power and a more gradual increase in LF power are observed In addition a strong increase in heart rate is observed immediately after standing up which aims to compensate the sudden decrease in blood pressure In the sample run this data file is analyzed by considering the lying and standing periods separately 5 1 Sample run In this sample run we show how to make the general analysis i e time domain frequency domain and nonlinear analysis for the lying and standing periods of the orthostatic mea surement separately This task can be easily accomplished in a single session First start the software and open the data file into the user interface At this point you can edit any of the analysis options to fit your demands If you are about to analyze several data files with the same options you better make these changes straight to the preferences The next thing to do is to select the RR interval
49. known probabilistic content 20 3 3 4 Detrended fluctuation analysis Detrended fluctuation analysis DFA measures the correlation within the signal The corre lation is extracted for different time scales as follows 36 First the RR interval time series is integrated A y k X RR RR k 1 N 3 15 j l where RR is the average RR interval Next the integrated series is divided into segments of equal length n Within each segment a least squares line is fitted into the data Let yn k denote these regression lines Next the integrated series y k is detrended by subtracting the local trend within each segment and the root mean square fluctuation of this integrated and detrended time series is calculated by N S ulk yn k 3 16 k 1 N This computation is repeated over different segment lengths to yield the index F n as a function of segment length n Typically F n increases with segment length A linear rela tionship on a double log graph indicates presence of fractal scaling and the fluctuations can be characterized by scaling exponent a the slope of the regression line relating log F n to logn Different values of a indicate the following a 1 5 Brown noise integral of white noise 1l lt a lt 1 5 Different kinds of noise a 1 1 f noise 0 5 lt a lt 1 Large values are likely to be followed by large value and vice versa a 0 5 white noise 0 lt a lt 0 5 Large value is likely to be followed by
50. lculated from the original non detrended RR series The Poincar plot and the DFA results are also presented graphically in the two axes In the Poincar plot left hand axis the successive RR intervals are plotted as blue circles and the SD1 and SD2 variables obtained from the ellipse fitting technique are presented for details see Section 3 3 1 In the DFA plot right hand axis the detrended fluctuations F n are presented as a function of n in a log log scale and the slopes for the short term and long term fluctuations a and ag respectively are indicated for details see Section 3 3 4 4 2 5 Menus and toolbar buttons The user menus and toolbar buttons are located on the upper left hand corner of the user interface There are all together three user menus and seven toolbar buttons The toolbar button icons and their actions are given below Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 2 The user interface 32 Nonlinear Analysis Results Variable Units Poincare plot SD1 ms sD2 ms Recurrence plot Mean line length Lmean beats Maximum fine length Lmax bests Recurrence rate REC Determinism DET Shannon Entropy ShanEn Other Approximate Entropy ApEn Sample Entropy SampEn Detrended fluctuations DFA a1 Detrended fluctuations DFA a2 Correlelation dimension D2 RR s dog yy N beats Calculated from the non de
51. led Poincar plot It is a graphical representation of the correlation between successive RR intervals i e plot of RR 1 as a function of RR as described in Fig 3 1 The shape of the plot is the essential feature A common approach to parameterize the shape is to fit an ellipse to the plot as shown in Fig 3 1 The ellipse is oriented according to the line of identity RR RR 1 6 The standard deviation of the points perpendicular to the line of identity denoted by SD1 describes short term variability which is mainly caused by RSA It can be shown that SD1 is related to the time domain measure SDSD according to 6 1 SD1 z5PSD 3 5 The standard deviation along the line of identity denoted by SD2 on the other hand de scribes long term variability and has been shown to be related to time domain measures SDNN and SDSD by 6 1 SD2 2SDNN 5SDSD 3 6 The standard Poincar plot can be considered to be of the first order The second order plot would be a three dimensional plot of values RR RR j 1 RR 2 In addition the lag can be bigger than 1 e g the plot RR RRj 2 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 3 Nonlinear methods 18 3 3 2 Approximate entropy Approximate entropy ApEn measures the complexity or irregularity of the signal 13 40 Large values of ApEn indicate high irregularity and smaller values
52. ll space of the regularization matrix Dg The null space of the second order difference matrix contains all first order curves and thus D2 is a good choice for estimating the aperiodic trend of RR series With these specific choices the detrended nearly stationary RR series can be written as Stat 2 HO I I MDI D2 z 2 6 In order to demonstrate the properties of the proposed detrending method its frequency response is considered Equation 2 5 can be written as 24a z where I I DTD corresponds to a time varying finite impulse response highpass filter The frequency response of for each discrete time point obtained as a Fourier transform of its rows is presented in Fig 2 4 a It can be seen that the filter is mostly constant but the beginning and end of the signal are handled differently The filtering effect is attenuated for the first and last elements of z and thus the distortion of end points of data is avoided The effect of the smoothing parameter on the frequency response of the filter is presented in Fig 2 4 b The cutoff frequency of the filter decreases when A is increased Besides the A parameter the frequency response naturally depends on the sampling rate of signal z Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND Chapter 3 Analysis methods In this chapter the analysis methods used in the soft
53. ls can be divided into technical and physiological artifacts The technical artifacts can include missing or additional QRS complex detections and errors in R wave occurrence times These artifacts may be due to measurement artifacts or the computational algorithm The physiological artifacts on the other hand include ectopic beats and arrhythmic events In order to avoid the interference of such artifacts the ECG recording and the corresponding event series should always be manually checked for artifacts and only artifact free sections should be included in the analysis 45 Alternatively if the amount of artifact free data is insufficient proper interpolation methods can be used to reduce these artifacts see e g 21 22 29 Another common feature that can alter the analysis significantly are the slow linear or more complex trends within the analyzed time series Such slow nonstationarities are characteristic for HRV signals and should be considered before the analysis The origins of nonstationarities in HRV are discussed e g in 3 Two kinds of methods have been used to get around the nonstationarity problem In 49 it was suggested that HRV data should be systematically tested for nonstationarities and that only stationary segments should be analyzed Representativeness of these segments in comparison with the whole HRV signal was however questioned in 15 Other methods try to remove the slow nonstationary trends from the HRV signal
54. lt overlap is 50 corresponding to 128 seconds In the AR spectrum there are also two options that can be selected First the order of the used AR model can be selected The default value for the model order is 16 but the model order should always be at least twice the number of spectral peaks in the data The second option is whether or not to use spectral factorization in the AR spectrum estimation In the fac torization the Ar spectrum is divided into separate components and the power estimates of each component are used for the band powers The factorization however has some serious problems which can distort the results significantly The main problems are the selection of the model order in such a way that only one AR component will result in each frequency band and secondly negative power values can result for closely spaced AR components Thus the selection of not to use factorization in AR spectrum is surely more robust and in that sense recommended 4 2 4 Results view The results for the selected RR interval sample are displayed in the results view segment The results are divided into time domain frequency domain and nonlinear results The results of each section are displayed by pressing the corresponding button on the top of the results view segment The results are by default updated automatically whenever any one of the the sample or analysis options that effect on the results is changed The updating of the results can be tim
55. ly as the powers of these components If factorization is disabled the AR spectrum powers are calculated as for the FFT spectrum The band powers in relative and normalized units are obtained from the absolute values as described in Table oul 3 3 Nonlinear methods Considering the complex control systems of the heart it is reasonable to assume that non linear mechanisms are involved in the genesis of HRV The nonlinear properties of HRV have been analyzed using measures such as Poincar plot 6 7 approximate and sample entropy 40 13 detrended fluctuation analysis 36 37 correlation dimension 16 18 and Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 3 Nonlinear methods 17 1000 F 950 F 900 F 850 F RR ms 800 F 750 F 700 F 650 F 650 700 750 800 850 900 950 1000 RR ms Figure 3 1 Poincar plot analysis with the ellipse fitting procedure SD1 and SD2 are the standard deviations in the directions x and x2 where xa is the line of identity for which RR RRj41 recurrence plots 48 47 50 During the last years the number of studies utilizing such methods have increased substantially The downside of these methods is still however the difficulty of physiological interpretation of the results 3 3 1 Poincar plot One commonly used nonlinear method that is simple to interpret is the so cal
56. mber of elements in the RP matrix for T l is equal to N m 1 and the recurrence rate is simply given as 1 N m 1 j k 1 The recurrence rate can also be calculated separately for each diagonal parallel to the line of identity main diagonal The trend of REC as a function of the time distance between these diagonals and the line of identity describes the fading of the recurrences for points further away The rest of the RP measures consider the lengths of the diagonal lines A threshold lmin 2 is used for excluding the diagonal lines formed by tangential motion of the trajectory The maximum line length is denoted Imax and its inverse the divergence 1 DIV 3 25 max has been shown to correlate with the largest positive Lyapunov exponent 47 The average diagonal line length on the other hand is obtained as Imax bain AI Imean 7 3 26 max N l Imin l Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 4 Summary of HRV parameters 23 where N is the number of length lines The determinism of the time series is measured by the variable Se 1 l lmin E a k Finally the Shannon information entropy of the line length distribution is defined as DET 3 27 lmax ShanEn 5 n ln n 3 28 l lmin where n is the number of length lines divided by the total number of lines that is N n 3 29 max
57. methods is given at the end of the chapter For most of the methods exact formulas for the different variables are given and possible parameter selections are pointed out In Chapter 4 the description of the features and usage of the software is given First the input data formats supported by the software are described and then the user interface through which the software is operated is described Then different options for saving the analysis results are described and finally instructions on how to set up the preference values for the analysis options are given In Chapter 5 two sample runs of the software are presented The first sample run describes how to analyze the lying and standing periods of the orthostatic test measurement distributed along this software separately as stationary segments The second sample run on the other hand describes the time varying analysis procedure of the same measurement In Appendix A some frequently asked questions with answers are given The questions are selected based on the feedback obtained from the users of the previous version of this software Finally in Appendix B workarounds for some commonly encountered technical problems are given Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND Chapter 2 Heart rate variability Heart rate variability HRV describes the variations between consecutive heartbeats Th
58. moothness priors detrending method can be compared to a time varying highpass filter and A adjusts the cut off frequency of the filter The estimated cut off frequency for a given lambda value is now presented next to the lambda value If you want to make sure that you don t eliminate any frequencies from the LF band you should select lambda in such a way that the corresponding cut off frequency is less than the lower limit of the LF band by default 0 04 Hz m How to select the interpolation rate of the RR series The interpolation rate is related to the cubic spline interpolation that is used for converting the RR interval series to equidistantly sampled series The default value for the interpolation rate is 4 Hz which works well for normal human HRV data It should be noted that the interpolation rate should be at least twice as high as the highest expected frequency in the RR interval series When changing the interpolation rate it should be remembered that it affects on the smoothness priors based detrending method i e when decreasing the interpolation rate also the A value of the smoothness priors method should be decreased 46 47 m How to select the window width and overlap for the FFT spectrum estimate The FFT spectrum in the software is calculated using the Welch s periodogram method where one or more overlapped segments are extracted from the data Then FFT spectrum is calculated for each segment and as a result the average
59. mponents The disadvantages of the AR spectrum are the complexity of model order selection and the contingency of negative components in the spectral factorization Nevertheless it may be advantageous to calculate the spectrum with both methods to have comparable results In this software the HRV spectrum is calculated with FFT based Welch s periodogram method and with the AR method Spectrum factorization in AR method is optional In the Welch s periodogram method the HRV sample is divided into overlapping segments The spectrum is then obtained by averaging the spectra of these segments This method decreases the variance of the FFT spectrum The generalized frequency bands in case of short term HRV recordings are the very low frequency VLF 0 0 04 Hz low frequency LF 0 04 0 15 Hz and high frequency HF 0 15 0 4 Hz The frequency domain measures extracted from the PSD estimate for each frequency band include absolute and relative powers of VLF LF and HF bands LF and HF band powers in normalized units the LF HF power ratio and peak frequencies for each band see Table 3 1 In the case of FFT spectrum absolute power values for each frequency band are obtained by simply integrating the spectrum over the band limits In the case of AR spectrum on the other hand if factorization is enabled distinct spectral components emerge for each frequency band with a proper selection of the model order and the absolute power values are obtained direct
60. n 2 0 home page on the web can be found at http kubios uku fi where you can find current information on the software and download possible updates and related material 1 5 Structure of this guide The aim of this guidebook is to help the user to get started with Kubios HRV It should not however be thought of as being an easy to follow step by step manual but more like a reference material from which you can probably find answers to your problems related to HRV analysis or usability of the software The structure of this guide is as follows After the overview chapter from where you will find useful information about the system requirements and installation an introduction to heart rate variability is given in Chapter 2 This chapter starts with a short discussion on the control systems of heart rate after which the extraction of heart beat periods is discussed and the derivation of HRV time series is Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 1 5 Structure of this guide 8 described The rest of the chapter is focused on the preprocessing of HRV data and gives a detailed description of the smoothness priors based detrending approach In Chapter 3 the analysis methods included in the software are described The de scriptions of the methods are divided into time domain frequency domain nonlinear and time varying categories and a summary of the
61. nal Analysis and Medical Imaging Group BSAMIG Uy version 2 0 Department of Physics University of Kuopio FINLAND Y 4 3 Saving the results 35 Print button opens a print dialog from which the report sheet can sent to the selected printer Export all pages to PDF file button is for exporting all report sheets into a single PDF file Zoom in button if for zooming in magnifying the report sheet Zoom out button is for zooming out the report sheet PLB D Reset to original size button can be used to restore the original zoom level This also resets the size of the corresponding report sheet window to its original size Move visible area button is for moving the visible area of the zoomed report sheet in the report window just grab the sheet with mouse and drag it to the desired direction Close button is for closing the report sheet The File menu includes Export All to PDF Print Current Page Print All Pages Close and Close All commands The Export All to PDF Print All Pages and Close commands are also given as toolbar buttons described above The last command Close All can be used for closing all report sheets simultaneously The Edit menu Windows version only contains Copy to Clipboard option which copies the contents of the corresponding report sheet window to the Windows clipboard This can be used to quickly copying the report sheet as an image into another program The Page menu includes commands for ch
62. om the Add or Remove Programs under the Windows Control Panel The uninstaller does not remove your preferences settings These have to be deleted manually from C Documents and Settings lt username gt Application Data KubiosHRV Manual uninstall The manual uninstallation should be conducted only if the automated uninstallation fails The Kubios HRV can be completely uninstalled manually by deleting the following files folders and registry entries e Delete the install folder by default C Program Files Kubios HRV and all the sub folders and files in it e Delete the KubiosHRV folder if exists from Windows XP 2000 C Documents and Settings lt username gt Application Data Windows Vista C Users lt username gt AppData Roaming e Delete the Kubios HRV quick launch icons if exist from Windows XP 2000 C Documents and Settings lt username gt Application Data Microsoft Internet Explorer Quick Launch Windows Vista C Users lt username gt AppData Roaming Microsoft Internet Explorer Quick Launch e Delete desktop icons if exist from Windows XP 2000 C Documents and Settings All Users Desktop Windows Vista C Users Public Desktop e Delete the possible Start menu entries from Windows XP 2000 C Documents and Settings All Users Start Menu Programs Windows Vista C ProgramData Microsoft Windows Start Menu Programs Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG ver
63. omputer B 1 Windows specific Kubios HRV Analysis fails to start and gives the following error message This application has failed to start because MCLMCRRT7x DLL was not found Re installing the application may fix this problem This error message is produced if the Kubios HRV executable cannot find the MATLAB Compiler Runtime MCR Verify that you have the MATLAB Compiler Runtime v7 x installed in lt KubiosHRV_install_dir gt MCR If you have the MCR installed and Kubios HRV Analysis still fails to start make sure that the system path contains the entry for MCR ie the lt KubiosHRV_install_dir gt MCR v7x runtime win32 directory is found in the system path Note that in order to modify the system path you need to have administrator privileges The quality of printed report sheets is poor with rasterized fonts etc Due to a bug in the MATLAB Compiler printing directly to a printer does not work in MATLAB created standalone applications in Windows platforms The workaround for this provided by MathWorks creates a bitmap file which it then sends to the printer The rasterization is a result of this bitmap con version There are however a few workarounds for this problem One ob vious workaround is to save the report sheet first as a PDF file and print it e g using Adobe Reader Another workaround is to install GhostScript and GSview http pages cs wisc edu ghost on your computer and use the 48 B 2
64. oted that the RR intervals must be measured stored in beat to beat In addition a support for plain RR interval ASCII text files is provided The input ASCII file can include either RR interval values or ECG data in one or two column format That is The RR interval values can be given as Type 1 Type 2 0 759 0 759 0 759 0 690 1 449 0 690 0 702 2 151 0 702 0 712 2 863 0 712 0 773 3 636 0 773 The RR interval values above are given in seconds but millisecond values can also be given 4 2 The user interface The developed HRV analysis software is operated with a graphical user interface This user interface window is shown in Fig 4 1 The user interface is divided into four segments 1 the RR interval series options segment on the top left corner 2 the data browser segment on the top right corner 3 the analysis options segment on the bottom left corner and 4 the results view segment on the bottom right corner Each of these segments are described in Sections 4 2 1 4 2 2 4 2 3 and 4 2 4 respectively 4 2 1 RR interval series options The RR interval series options shown in Fig 4 2 include three functions Artifact correction Samples for analysis and Remove trend components The artifact correction options can be 25 4 2 The user interface 26 Kubios HRV Analysis V Matlab HRV_v2_matlab data suunto_data sdf Sample 1 Length h min s Method Smoothn priors Lambda 500 f 0 035Hz 00 03 20 00 05 00 00 1
65. pathy Diabetes Nutri tion amp Metabolism 13 6 341 346 2000 O Pahlm and L S rnmo Software QRS detection in ambulatory monitoring a review Med Biol Eng Comput 22 289 297 July 1984 J Pan and W J Tompkins A real time QRS detection algorithm IEEE Trans Biomed Eng 32 3 230 236 March 1985 C K Peng S Havlin H E Stanley and A L Goldberger Quantification of scaling exponents and crossover phenomena in nonstationary heartbeat time series Chaos 5 82 87 1995 T Penzel J W Kantelhardt L Grote J H Peter and A Bunde Comparison of detrended fluctuation analysis and spectral analysis for heart rate variability in sleep and sleep apnea IEEE Trans Biomed Eng 50 10 1143 1151 October 2003 G D Pinna R Maestri A Di Cesare R Colombo and G Minuco The accuracy of power spectrum analysis of heart rate variability from annotated RR lists generated by Holter systems Physiol Meas 15 163 179 1994 S W Porges and R E Bohrer The analysis of periodic processes in psychophysiological research In J T Cacioppo and L G Tassinary editors Principles of Psychophysiology Physical Social and Inferential Elements pages 708 753 Cambridge University Press 1990 J A Richman and J R Moorman Physiological time series analysis using approximate entropy and sample entropy Am J Physiol 278 H2039 H2049 2000 O Rompelman Rhythms and analysis techniques In J Strackee and N Wester hof editors The Phys
66. raw RR interval series as can be seen from Fig 4 3 a The correction level can be accepted and the correction performed by pressing the Apply button The corrected series is shown in Fig 4 3 b As can be seen from the spectrum estimates in Figs 4 3 a and b the correction of just few artifacts has a very significant effect on the analysis results Thus even single artifacts should always be taken care of prior to HRV analysis In the Samples for analysis options the part s of the RR interval series to be analyzed can be modified by adding or removing samples and by changing the start time or length of po Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG y SF version 2 0 Department of Physics University of Kuopio FINLAND Y 4 2 The user interface 27 00 00 24 00 05 o0 Remove trend components Method Smoothn priors i Lambda 500 f 0 035 Hz Figure 4 2 The RR interval series options segment of the user interface the sample If more than one sample is selected the analysis can be done either for the single samples separately or by merging the samples into one long sample before analysis This selection is visible under the RR series axis when multiple samples are selected The starting point and length of the samples can also be changed by moving resizing the patch over the RR series as described in Section 4 2 2 This section also describes how to add remove samples to from RR series axes
67. re the spectrum can not be considered to be a function of frequency but rather of cycles per beat 10 Another common approach adopted in this software is to use interpolation methods for converting the non equidistantly sampled RR interval time series also called the interval function to equidistantly sampled 45 see the right panel of Fig 2 3 One choice for the interpolation method is the cubic spline interpolation 28 After interpolation regular spectrum estimation methods can be applied The third general approach called the spectrum of counts considers a series of impulses delta functions positioned at beat occurrence times 11 This approach relies on the generally accepted integral pulse frequency modulator IPFM which aims to model the neural modulation of the SA node 41 According to this model the modulating signal is integrated until a reference level is achieved after which an impulse is emitted and the integrator is set to zero The spectrum of the series of events can be calculated e g by first lowpass filtering the event series and then calculating the spectrum of the resulting signal 10 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 2 3 Preprocessing of HRV time series 13 2 3 Preprocessing of HRV time series Any artifact in the RR interval time series may interfere the analysis of these signals The artifacts within HRV signa
68. rical N m 1 7 x N m 1 7 matrix of zeros and ones The element in the j th row and k th column of the RP matrix i e RP j k is 1 if the point uj on the trajectory is close to point uz That is 1 d uj uk lt r RP A d 0 otherwise Geo where d u ux is the Euclidean distance given in 3 19 and r is a fixed threshold The structure of the RP matrix usually shows short line segments of ones parallel to the main diagonal The lengths of these diagonal lines describe the duration of which the two points are close to each other An example RP for HRV time series is presented in Fig 3 4 Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 3 3 Nonlinear methods 22 AAA bl Enhi 4 Time min Figure 3 4 Recurrence plot matrix for HRV time series black 1 and white 0 Methods for quantifying recurrence plots were proposed in 48 The methods included in this software are introduced below In the software the following selections were made The embedding dimension and lag were selected to be m 10 and 7 1 respectively The threshold distance r was selected to be ym SD where SD is the standard deviation of the RR time series The selection are similar to those made in 8 The first quantitative measure of RP is the recurrence rate REC which is simply the ratio of ones and zeros in the RP matrix The nu
69. series analysis In M Akay editor Nonlinear Biomedical Signal Processing Dynamic Analysis and Mod eling volume II chapter 1 pages 1 39 IEEE Press New York 2001 H V Huikuri T H Makikallio P Raatikainen J Perki m ki A Castellanos and R J Myerburg Prediction of sudden cardiac death appraisal of the studies and methods assessing the risk of sudden arrhythmic death Circulation 108 1 110 115 July 2003 D E Lake J S Richman M P Griffin and J R Moorman Sample entropy analysis of neonatal heart rate variability ajp 283 R789 R797 September 2002 N Lippman K M Stein and B B Lerman Nonlinear predictive interpolation a new method for the correction of ectopic beats for heart rate variability analysis J Electrocardiol 26 514 519 1993 N Lippman K M Stein and B B Lerman Comparison of methods for removal of ectopy in measurement of heart rate variability Am J Physiol 267 1 H411 H418 July 1994 D A Litvack T F Oberlander L H Carney and J P Saul Time and frequency domain methods for heart rate variability analysis a methodological comparison Psychophys iol 32 492 504 1995 F Lombardi T H Makikallio R J Myerburg and H Huikuri Sudden cardiac death role of heart rate variability to identify patients at risk Cardiovasc Res 50 210 217 2001 A Malliani M Pagani F Lombardi and S Cerutti Cardiovascular neural regulation explored in the frequency domain Circulation 84 2 4
70. sion 2 0 Department of Physics University of Kuopio FINLAND 1 4 Software home page 7 e Remove the MATLAB Compiler Runtime entry lt KubiosHRV install dir gt MCR v78 runtime win32 from the system path e Manual removal of the Kubios HRV Analysis entry from the Windows Add or Re move Programs list requires modifying registry A thorough instructions on how to manually remove programs from the Add or Remove Programs list is available on the Microsoft support web site at http support microsoft com kbid 314481 1 3 2 Linux Automated uninstall Open terminal and change directory to the Kubios HRV install directory Run the command sh uninstall_kubios sh and follow instructions given on screen NOTE You may need root privileges for running the uninstaller if you have installed Kubios HRV as root Manual uninstall Open a terminal and run the following commands In the commands replace lt Kubios HRV install dir gt with the Kubios HRV install directory NOTE You may need root privileges for running these commands if you have installed Kubios HRV as root e Uninstall Kubios HRV menu entries xdg desktop menu uninstall lt Kubios HRV install dir gt BSAMIG KubiosHRV desktop e Uninstall Kubios HRV desktop icons xdg desktop icon uninstall lt Kubios HRV install dir gt BSAMIG KubiosHRV desktop e Delete the Kubios HRV install directory rm rf lt Kubios HRV install dir gt 1 4 Software home page The Kubios HRV versio
71. tion Analysis options User Details Name Mika Tarvainen Department Department of Physics Advanced settings Organization University of Kuopio These details wil be displayed in the report Sheet and ASCII results fle Report settings Figure 4 10 Set up preferences window of the software User information category Param The analysis options used in the calculation of the results Data The RR interval data Statistics Time domain analysis results Frequency Frequency domain analysis results NonLinear Nonlinear analysis results The variable names of the different fields are more or less self descriptive and are not docu mented here 4 4 Setting up the preferences All the analysis options that can be adjusted in the user interface have some default values These preference values will be used every time the program is started Any changes made on these values in the user interface only apply for the current session The preference values are designed to be more or less suitable for short term HRV recordings and may sometimes need to be redefined This can be done by selecting Edit Preferences from the File menu or by pressing the corresponding toolbar button This will open the preferences window in which the preference values can be redefined The preferences are divided into four categories User information Analysis options Advanced settings and Report settings In the User information category sho
72. trended selected RR series Figure 4 8 The results view segment of the user interface nonlinear results view selected Open new data file button is for opening a new data file for anal ysis If the results of the current analysis have not been saved user is prompted to do so Save results button is for saving the analysis results The results can be saved in ASCII PDF and MATLAB MAT file format see Section 4 3 for details Print results button is for printing the current results without open ing report sheet windows Report sheet button opens one or several report sheet windows which include all the analysis results see Section 4 3 2 for details Edit preferences button opens a preferences window in which you can e g change the default values for analysis options see Section 4 4 for details About HRV analysis software button opens the about dialog of the software which includes the version number and contact infor mation Also the Kubios HRV End User License Agreement can be viewed in the about dialog Open Kubios HRV User s Guide button opens the Kubios HRV User s Guide this document PDF file using the default PDF viewer of the system Close file button closes the current data file If the results of the current analysis have not been saved user is prompted to do so Co 882 0 0 All the above actions are also available on the user menus The File menu includes Open Save Results Save Results As
73. ullPointerException null attribute This error is due to a bug in the communication between Java and CUPS Com mon Unix Printing System 1 3 4 The possible workarounds for this issue are e Use the custom print command options in the Kubios HRV preferences e Set the orientation property to anything else than Automatic for every printer installed on the system e g in Ubuntu Linux this can be done by opening the printer configuration tool System Administration Printing and changing the orientation property from the Job Options tab for every installed printer e Downgrade you CUPS to older version not recommended Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND References 1 2 V X Afonso ECG QRS detection In W J Tompkins editor Biomedical Digital Signal Processing chapter 12 pages 237 264 Prentice Hall New Jersey 1993 G Baselli S Cerutti S Civardi F Lombardi A Malliani M Merri M Pagani and G Rizzo Heart rate variability signal processing a quantitative approach as an aid to diagnosis in cardiovascular pathologies Int J Bio Med Comput 20 51 70 1987 G G Berntson J T Bigger Jr D L Eckberg P Grossman P G Kaufmann M Malik H N Nagaraja S W Porges J P Saul P H Stone and M W Van Der Molen Heart rate variability Origins methods and interpretive caveats Psychophysiol 34
74. val histogram divided by the height of the histogram 45 Baseline width of the RR interval histogram 45 VLF LF and HF band peak frequencies Absolute powers of VLF LF and HF bands Relative powers of VLF LF and HF bands VLF VLF ms total power ms x 100 LF LF ms total power ms x 100 HF HF ms total power ms x 100 Powers of LF and HF bands in normalized units LF n u LF ms total power ms VLF ms HF n u HF ms total power ms VLF ms Ratio between LF and HF band powers The standard deviation of the Poincar plot perpendicular to SD1 and along SD2 the line of identity 6 7 Approximate entropy Eq 3 11 40 13 Sample entropy Eq 3 14 40 Correlation dimension Eq 3 21 16 18 Detrended fluctuation analysis 36 37 Short term fluctuation slope Long term fluctuation slope Recurrence plot analysis 48 8 50 Mean line length Eq 3 26 Maximum line length Recurrence rate Eq 3 24 Determinism Eq 3 27 Shannon entropy Eq 3 28 Biosignal Analysis and Medical Imaging Group BSAMIG Department of Physics University of Kuopio FINLAND Chapter 4 Software description 4 1 Input data formats Kubios HRV supports the following RR interval file formats First of all data of two com monly used heart rate monitors is supported These are SUUNTO SDF STE and POLAR HRM files When analyzing data of these devices it should however be n
75. w the instructions given in the setup wizard to complete installation You can launch the Kubios HRV by selecting it from the created Start Menu folder or by clicking the Desktop icon if created Please note that the starting of Kubios HRV also starts the MATLAB Compiler Runtime and may take some time especially with older computers 1 2 2 Linux Run the Kubios HRV installer by typing sh KubiosHRV linux 2 0x86 run in the ter minal and follow the instructions given on screen The Kubios HRV installer also includes the MATLAB Compiler Runtime NOTE If you run the installer as root Kubios HRV will be installed on your computer system wide for all users If you want to install Kubios HRV only for yourself run the installer as local user Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 1 3 Uninstallation 6 1 3 Uninstallation The Kubios HRV can be uninstalled either automatically using the uninstaller or manually if the uninstaller fails for some reason Both methods for uninstallation are described in the following 1 3 1 Windows Automated uninstall The preferred and the most straightforward way of uninstalling Kubios HRV is to use the automated uninstaller The uninstaller can be launched by selecting Uninstall Kubios HRV from the software s Start menu folder the default Start menu folder is Kubios HRV The software can also be uninstalled fr
76. ware are introduced The presented methods are mainly based on the guidelines given in 45 The presentation of the methods is divided into three categories i e time domain frequency domain and nonlinear methods The methods summarized in Table 3 1 3 1 Time domain methods The time domain methods are the simplest to perform since they are applied straight to the series of successive RR interval values The most evident such measure is the mean value of RR intervals RR or correspondingly the mean HR HR In addition several variables that measure the variability within the RR series exist The standard deviation of RR intervals SDNN is defined as SDNN 3 1 where RR denotes the value of 7 th RR interval and N is the total number of successive intervals The SDNN reflects the overall both short term and long term variation within the RR interval series whereas the standard deviation of successive RR interval differences SDSD given by SDSD y E ARR5 E ARR 3 2 can be used as a measure of the short term variability For stationary RR series E ARR E RRj41 E RR 0 and SDSD equals the root mean square of successive differences RMSSD given by 1 N 1 RMSSD 2 RRjy1 RR 3 3 Another measure calculated from successive RR interval differences is the NN50 which is the number of successive intervals differing more than 50 ms or the corresponding relative amount NN50 N 1
77. wn in Fig 4 10 you can set up your personal contact information Name Department and Organization This information will only be included in the bottom left corner of the report sheet and in the beginning of the ASCII text file including the analysis results That is the user information is meant just for indicating the person who has carried out the analysis jh Kubios HRV Biosignal Analysis and Medical Imaging Group BSAMIG version 2 0 Department of Physics University of Kuopio FINLAND 4 4 Setting up the preferences 37 Preferences Analysis options User information Default Input Data Type Advanced settings E Number of samples 1 Sample analysis type Single samples RR Interval Detrending Detrending method smoothn ps smoothing Parameter oo HRV Frequency Bands Report settings Very low frequency VLF o 0 04 Low frequency LF 0 04 015 High frequency HF 015 o4 Update Analysis Results Update mode Figure 4 11 Set up preferences window of the software Analysis options category The Analysis options category shown in Fig 4 11 includes some basic analysis op tions The default input data type can be set to one of the file formats mentioned in Section 4 1 and the selected data type will be used as default every time a new data file is opened In addition the analysis options category includes RR interval samples RR interval de trending HRV frequency bands and Up
78. z HF 015 0442 Max fine tengin Lmax Recurrence rate REC Determinism DET Shamon Entropy SnanEn Other ApEn SampEn Detrended tuotations DFA at Detrended Suotustions DFA 02 Corretstion dimension D2 Department of Physics University of Kuopio 8 PSD wg VEF 9004 52 317 UF 0 04 0 15 Hz HF 015 04 Hz Total 2103 LEHE 0322 Poincare Plot Detrended fluctuations DFA 12 13 06 08 1 12 14 16 18 RR 5 tog n De ts Results are calculated trom the nor Cetrenced selectes AR series Kubios HRV Analysis version 2 0 Department of Physics University of Kuopio Finland Figure 5 2 Sample run 1 results for the lying period of the orthostatic test Kubios HRV y version 2 0 i Biosignal Analysis and Medical Imaging Group BSAMIG Department of Physics University of Kuopio FINLAND 5 1 Sample run Report Page 2 File Edit SAA RI NRAAPO HRV Analysis Results RR Interval Time Series 00 00 00 00 01 40 Selected Detrended RR Series 01 Time Domain Results Variable Uns Mean RAC ms STD RR SONN ms Mean HA 1min STOHR iminy ms court RR sianguisr index N TIN ms Frequency Domain Results FFT spectrum WWelon s periocogram 256 s window win 50 overlap Frequency VLF 0 008 Hz LF 0 04 0 15 H2 HF 01504 Hz Recurrence rate REC Determinism DET Shannon Entropy SrarEn Other Appraamae entropy ADEN Sampie entropy

Download Pdf Manuals

Related Search

Related Contents

Page 1 Page 2 このたびはカワイ電子ピアノをお求めい ~ただきま して  GloBrite Manual 802012 REV C 2-11-13  operating instructions  仕 様 書 1 件名 平成 23 年度埼玉地域支社 放射線量測定装置の購入 2  SCHÉMA DE COHÉRENCE TERRITORIALE (SCOT)  Siemens S223 User's Manual  L:7/"le/`7  Whirlpool 2195385 Refrigerator User Manual  User Manual  Manual - Omega Bio-Tek  

Copyright © All rights reserved. Failed to retrieve file