Blood pressure waveform analysis by means of wavelet transform

The assessment of cardiovascular function by means of arterial pulse wave analysis (PWA) is well established in clinical practice. PWA is applied to study risk stratification in hypertension, with emphasis on the measurement of the augmentation index as a measure of aortic pressure wave reflections. Despite the fact that the prognostic power of PWA, in its current form, still remains to be demonstrated in the general population, there is general agreement that analysis and interpretation of the waveform might provide a deeper insight in cardiovascular pathophysiology. We propose here the use of wavelet analysis (WA) as a tool to quantify arterial pressure waveform features, with a twofold aim. First, we discuss a specific use of wavelet transform in the study of pressure waveform morphology, and its potential role in ascertaining the dynamics of temporal properties of arterial pressure waveforms. Second, we apply WA to evaluate a database of carotid artery pressure waveforms of healthy middle-aged women and men. Wavelet analysis has the potential to extract specific features (wavelet details), related to wave reflection and aortic valve closure, from a measured waveform. Analysis showed that the fifth detail, one of the waveform features extracted applying the wavelet decomposition, appeared to be the most appropriate for the analysis of carotid artery pressure waveforms. What remains to be assessed is how the information embedded in this detail can be further processed and transformed into quantitative data, and how it can be rendered useful for automated waveform classification and arterial function parameters with potential clinical applications.     

Finger and ear photoplethysmogram waveform analysis by fitting with Gaussians

Analysis of the contour of the blood volume pulse (VP) has become important because it contains much information about cardiovascular activity. Traditionally, pulse contour analysis requires first or higher derivatives to be calculated. This paper describes a novel algorithm for analysing simultaneously measured ear and finger photoplethysmography (PPG) signals. The algorithm separates the systolic wave and the diastolic wave of the VP and fits each of them with the sum of two Gaussian functions. The VP was obtained from PPG signals taken from 40 healthy subjects at each heartbeat cycle. From the evaluated VP, time values of the direct wave and three reflected waves were calculated, as well as the augmentation index (AI) and the reflection index (RI). The evaluated parameters were compared with those that were obtained by the derivative method, and it was demonstrated that the new method can be used to analyze VP waveforms.     

First in vivo application and evaluation of a novel method for non-invasive estimation of cardiac output

Surgical or critically ill patients often require continuous assessment of cardiac output (CO) for diagnostic purposes or for guiding therapeutic interventions. A new method of non-invasive CO estimation has been recently developed, which is based on pressure wave analysis. However, its validity has been examined only in silico. Aim of this study was to evaluate in vivo the reproducibility and accuracy of the “systolic volume balance” method (SVB). Twenty two subjects underwent 2-D transthoracic echocardiography for CO measurement (reference value of CO). The application of SVB method required aortic pressure wave analysis and estimation of total arterial compliance. Aortic pulses were derived by mathematical transformation of radial pressure waves recorded by applanation tonometry. Total compliance was estimated by the “pulse pressure” method. The agreement, association, variability, bias and precision between Doppler and SVB measures of CO were evaluated by intraclass correlation coefficient (ICC), mean difference, SD of differences, percentage error (PR) and Bland–Altman analysis. SVB yielded very reproducible CO estimates (ICC = 0.84, mean difference 0.27 ± 0.73 L/min, PR = 16.7%). SVB-derived CO was comparable with Doppler measurements, indicating a good agreement and accuracy (ICC = 0.74, mean difference = −0.22 ± 0.364 L/min, PR ≈ 15). The basic mathematical and physical principles of the SVB method provide highly reproducible and accurate estimates of CO compared with echocardiography.