Increasing Stability of Real-Time Pulse Wave Velocity Estimation by Combining Established and New Approaches

Abstract

Pulse wave velocity (PWV) in arteries is an estimator for arterial stiffness. As velocity is calculated as distance per time, the accuracy and reliability of PWV measurements depends heavily on the estimation of the pulse transit time (PTT). Several methods for the estimation of the PTT exist, often resulting in different PWV values. They rely on the exact determination of specific points in the pulse wave, which are easily affected by small changes of the signal. As there is still no agreement on the accuracy of these methods, only the reliability and stability of these commonly used "foot-to-foot" algorithms are compared in this paper. These algorithms are based on the detection of the diastole-minimum, the maximum of the first or second derivative or combinations thereof. We also adapted a new approach called "diastole-patching" based on the matching of a specific region of the pulse waves. The methods were tested using 2348 pulse waves from the MIMIC Database, collected from 46 subjects. Intra-subject deviation and relative outliers per subject were lowest in the adapted diastole-patching algorithm (4.9 ± 3ms and 0.6 ± 2.7%, respectively). Therefore, this study has shown that the diastole-patching method is the most stable and reliable of the algorithms under investigation.