Reservoir Analysis using a matlab batch process for multiple Sphygmocor© files (bRes_sp)
For background see Alun Hughes, Kim Parker. The modified arterial reservoir: an update with consideration of asymptotic pressure (P∞) and zero-flow pressure (Pzf). Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine in press. https://doi.org/10.1177/0954411920917557 and Kim Parker's website pages on Reservoir/excess pressure (http://www.bg.ic.ac.uk/research/k.parker/res_press_web/rp_home.html).
I am grateful to my long-term colleague Prof. Kim Parker for his assistance with the development of this program.
This MATLAB script replaces a previous version called batch_res. Earlier versions of batch_res and kreservoir (v13) may be available on Kim Parker's web page. NB that old versions (batch_res_vXX and kreservoir_v13) are now outdated and the current version (bRes_sp and kreservoir v14) should be used. Henceforth any new versions will be made available here.
| Version | Date | Changes Made |
|---|---|---|
| 1.1 | 24/12/17 | Added information about program upgrade to kreservoir v9 that includes HRV |
| 1.2 | 27/12/17 | Incorporated estimates of wave intensity, backward pressure (Pb) and forward pressure (Pf) from central pulse waveform and used sgolayfilt rather than fsg1521 or fsg721 |
| 1.21 | 22/10/18 | Minor updates to accompany a few bug fixes (now kreservoir v11) |
| 1.3 | 17/03/19 | Improved reservoir algorithm to prevent upturn of pressure at end of diastole (early systole) affecting fit. Also improved HRV algorithm; adding data to excel output, fixing some other bugs (now kreservoir v12) |
| 1.31 | 12/04/19 | Added progress bar, bug fix to figure output (now kreservoir v13) |
| 1.4 | 11/01/20 | Some restructuring of functions, minor bug fixes (now kreservoir v14) |
| 1.41 | 11/05/20 | More error traps, revised to use textscan rather than textread, titles for figures improved peak detection for Wf2 |
| 1.5 | 23/05/20 | Updated script to calculate derivatives using 9 frame 3rd order SG filter and renamed program to bRes_sp. |
| 1.6 | 25/03/21 | Fixed a couple of bugs with findpeaks component |
Heart rate variability (HRV) and baroreceptor sensitivity measures
bRes_sp is a matlab script that calculates reservoir and excess pressure according to the methods described in Davies et al.[1] for Sphygmocor© derived files. A few minor changes have been made since the original description. An improved algorithm for fitting the reservoir in diastole is now used -- this excludes upstrokes at the end of diastole from the fit (presumed to be due to the next beat). This results in lower values for P∞ and slightly different values for other reservoir parameters. The program assumes that the first element of the pressure P corresponds to the end diastolic pressure in the arterial pressure waveform; i.e. the time of minimum P just before the rapid rise in P during early systole.
Put sphygmocor files to be analysed in the analysis directory
C:\Spdata
Open matlab and ensure that the working directory is the one that contains the relevant script and function files (in my case this is C:\XXX\Matlab\work\Reservoir_batch).
Type into command line:
>>bRes_sp
After some time (depending on how many files are analysed the run should complete, returning to command prompt. It will show a progress bar while it is running.
Two new folders should now exist in C:\Spdata - C:\Spdata\figures and C:\Spdata\results.
C:\Spdata\figures contains figures of the signal (all the uncalibrated data) and the pulse (the calibrated ensemble averaged data) with the reservoir pressure shown. These are saved as *.jpg files.These plots are useful for checking quality and any dubious results.
C:\Spdata\results will contain an excel file (resdata.xls) which will contain all the data for each file with its ID. This can then be imported into Stata (or some other stats program) for further analysis.
[NB THESE MEASURES ARE EXPERIMENTAL FOR SPHYGMOCOR DATA]
The root mean square of successive differences (RMSSD), the standard deviation of the pulse intervals (SDNN) and baroreflex sensitivity (BRS) are calculated essentially according to Sluyter et al.[2],[a] The validity of such ultrashort recordings has been studied by Munoz et al.[3] Further details on the meaning and interpretation of these measures can be found in Shaffer and Ginsberg.[4] It is probably useful to normalise HRV (or adjust it statistically) to mean RR interval due to the correlation between HRV and resting heart rate.[5] This can be done as a post-processing step in the statistical package used.
[NB THESE MEASURES ARE EXPERIMENTAL FOR SPHYGMOCOR DATA]
If it is assumed that excess pressure (Pxs) is proportional to aortic flow velocity (U) (essentially a 3-element Windkessel assumption -- see above) then the pattern of aortic wave intensity (dI) can be estimated (being proportional to dP x dPxs). If one of aortic wave speed or dU is known then wave intensity can be estimated. If only pressure has been measured this problem cannot be solved without strong assumptions. In this case, it is assumed that peak aortic flow (dU) is 1m/s (based on [6]) and doesn't not vary with age, sex etc. While this is not true, it is may prove an acceptable approximation. Further details including preliminary validation can be found in [7].
[NB THESE MEASURES ARE EXPERIMENTAL FOR SPHYGMOCOR DATA]
These are calculated based on the assumptions that in the aorta reservoir pressure is 2 x backward pressure (Pb);[8] which may be valid if excess pressure is linearly proportional to aortic flow as has been reported in dogs,[9] and total aortic flow equals aortic inflow. This approach probably shares similarities with the ARCSOLVER method,[10] which uses a 3-element Windkessel assumption[b] to reconstruct forward and backward pressures.
| Variable | Definition | Example result | Units |
|---|---|---|---|
| re_file | File identifier | xxx_pwa.txt | No units |
| re_maxp | Maximum pressure (systolic pressure) | 120 | mmHg |
| re_tmaxp | Time of maximum (systolic) pressure | 0.1484375 | s |
| re_minp | Minimum pressure (diastolic pressure) | 76.55 | mmHg |
| re_intpr | Integral of reservoir pressure | 112.3308474 | mmHg.s |
| re_maxpr | Maximum reservoir pressure | 108.1486734 | mmHg |
| re_tmaxpr | Time of maximum reservoir pressure | 0.265625 | s |
| re_intprlessdias | Integral of reservoir pressure with diastolic pressure subtracted | 16.09752936 | mmHg.s |
| re_maxprlessdias | Maximum of reservoir pressure with diastolic pressure subtracted | 31.64019694 | mmHg |
| re_sam_rate | Sampling rate | 128 | Hz |
| re_intxsp | Integral excess pressure | 4.169230704 | mmHg.s |
| re_maxxsp | Maximum excess pressure | 26.3460721 | mmHg |
| re_tmaxxsp | Time of maximum excess pressure | 0.109375 | s |
| re_tn | Time of maximum -dp/dt (nominal end of systole) | 0.283007813 | s |
| re_pinf | P∞ | 70.70521706 | mmHg |
| re_pn | Pressure at start of diastole | 110.95 | mmHg |
| re_fita | Rate constant systolic fit | 10.45715136 | s-1 |
| re_fitb | Rate constant diastolic fit | 1.912785514 | s-1 |
| re_rsq | Coefficient of determination (r^2^) for fit | 0.992749217 | No units |
| re_prob | Flag 1 for likely problem[c] | 0 | No units |
| re_version | kreservoir version (for version tracking | v13 | No units |
| re_sdsbp_mmhg | Standard deviation of SBP | 6.1 | mmHg |
| re_rr_interval | Pulse to pulse (RR) interval | 900 | ms |
| re_rmssd | Root mean square of differences in successive pulse (RR) intervals | 10 | ms |
| re_ssdn | Standard deviation of pulse intervals | 6 | ms |
| re_brs | Baroreflex sensitivity (BRS) by the sequence method | 22 | ms.mmHg-1 |
| re_brs_valid | Number of valid BRS measures | 8 | Count |
| re_pb_pf | Central Pb/Pf | 0.65 | No units |
| re_ri | Central Reflection index | .4 | No units |
| re_wf1i | Intensity of forward compression wave 1 (W1) | W/m2* | |
| re_wf1t | Time of peak of forward compression wave 1 (W1) | s | |
| re_wf1a | Area of forward compression wave 1 (W1) | J/m2* | |
| re_wfbi | Intensity of backward compression wave (Wb) | W/m2* | |
| re_wbt | Time of peak of backward compression wave (Wb) | s | |
| re_wba | Area of backward compression wave (Wb) | J/m2* | |
| re_wf2i | Intensity of forward compression wave 2 (W2) | W/m2* | |
| re_wf2t | Time of peak of forward compression wave 2 (W2) | s | |
| re_wf2a | Area of forward compression wave 2 (W2) | J/m2* | |
| wri | Wave reflection index | No units | |
| rhoc | Wave speed | m/s |
* if calibrated.
- Davies JE, Lacy P, Tillin T, et al. Excess pressure integral predicts cardiovascular events independent of other risk factors in the conduit artery functional evaluation substudy of Anglo-Scandinavian Cardiac Outcomes Trial. Hypertension 2014; 64(1): 60-8.
- Sluyter JD, Hughes AD, Camargo CA, Jr., Lowe A, Scragg RKR. Relations of Demographic and Clinical Factors With Cardiovascular Autonomic Function in a Population-Based Study: An Assessment By Quantile Regression. Am J Hypertens 2017; 31(1): 53-62.
- Munoz ML, van Roon A, Riese H, et al. Validity of (Ultra-)Short Recordings for Heart Rate Variability Measurements. PLoS One 2015; 10(9): e0138921.
- Shaffer F, Ginsberg JP. An Overview of Heart Rate Variability Metrics and Norms. Front Public Health 2017; 5: 258.
- van Roon AM, Snieder H, Lefrandt JD, de Geus EJ, Riese H. Parsimonious Correction of Heart Rate Variability for Its Dependency on Heart Rate. Hypertension 2016; 68(5): e63-e5.
- Hughes AD, Park C, Ramakrishnan A, Mayet J, Chaturvedi N and Parker KH. Feasibility of Estimation of Aortic Wave Intensity Using Non-invasive Pressure Recordings in the Absence of Flow Velocity in Man. Front Physiol 2020; 11:550. doi: 10.3389/fphys.2020.00550
- Hughes AD, Park C, Ramakrishnan A, Mayet J, Chaturvedi N, Parker KH. Feasibility of Estimation of Aortic Wave Intensity Using Non-invasive Pressure Recordings in the Absence of Flow Velocity in Man. Front Physiol 2020; 11: in press.
- Westerhof N, Westerhof BE. The reservoir wave paradigm discussion. J Hypertens 2015; 33(3): 458-60.
- Wang J, Jr., O'Brien AB, Shrive NG, Parker KH, Tyberg JV. Time-domain representation of ventricular-arterial coupling as a windkessel and wave system. Am J Physiol Heart Circ Physiol 2003; 284(4): H1358--68.
- Hametner B, Wassertheurer S, Kropf J, et al. Wave reflection quantification based on pressure waveforms alone--methods, comparison, and clinical covariates. Comput Meth Prog Bio 2013; 109(3): 250-9.
[a]: since Sphygmocor data is cropped at the foot of the waveform peak systole has been used as the fiducial point of the waveform to calculate beat to beat intervals.
[b]: the details of the procedure used by ARCSOLVER are not in the public domain
[c]: 0 = ok; 1 = Pinf > diastolic pressure; 2 = rate constant b < 0; 3 = time of maximum reservoir pressure > end of systole