| PhD Viva

Name of the Speaker: Mr. Rahul M (EE17D202)
Guide: Prof. Mohanasankar Sivaprakasam
Co-Guide: Dr. Jayaraj Joseph
Online meeting link: https://meet.google.com/sjb-kndt-zxb
Date/Time: 28th June 2024 (Friday), 3:00 PM
Title: Demystifying the arterial pulse wave: a system and method to quantify pulse wave reflections and its novel applications

Abstract :

Early detection and timely intervention are critical for reducing mortality, minimising disease burden, and enhancing outcomes in cardiovascular diseases (CVD). Traditional risk factors fall short in detecting subclinical CVD, whereas early vascular ageing (EVA) markers derived from arterial pulse waves detect the early manifestations of CVD. The arterial pulse wave, a signature of vascular health, is moulded by pulse wave reflections (PWR) in the arterial system. Quantification of PWR elucidates physiologic mechanisms of EVA and is pivotal for reliable assessment of local pulse wave velocity (PWV) – an emerging EVA marker. Current methods to quantify PWR using wave separation analysis (WSA) require simultaneous pressure (P) and flow (Q) from the aorta, posing challenges due to the need for non-invasive techniques and complex instrumentation. This PhD work addresses these challenges by proposing bio-physics models and instrumentation for quantifying PWR using a single pulse wave and its novel applications in improving the reliability of measuring the local PWV.

The arterial pulse wave was decomposed using a Multi-Gaussian Decomposition Model (MGDM) of a pulse waveform with weighted and shifted Gaussians for WSA. The wave separation was obtained using empirically derived relationships and ground-truth parametric analysis. In the second approach, Q was constructed as a Multi-Rayleigh Flow Model using weighted and shifted Rayleigh functions. The model parameters were derived from P and analysed in frequency domain for WSA.

The performance of these methods was validated on in-silico virtual subjects database (N = 4374, healthy, age: 25 – 75 years) and in-vivo studies on human participants (N = 70, healthy, age: 20 – 51 years). Reliable forward and backward pulse waves were obtained using both approaches, with root-mean-square error of 2.5 mmHg. The reflection quantification indices obtained from both methods had a statistically significant and strong correlation (r > 0.75, p < 0.001) with those obtained using reference methods.

Controlled experiments and evidence from literature, suggests that PWR corrupts the pulse morphology in the systolic phase, affecting the reliability of measuring local PWV. The WSA based on the proposed methods was used to improve the reliability and repeatability of measuring local PWV at the common carotid artery. A measurement apparatus consisting of a dual-channel ultrasound transducer with an integrated pressure sensor was developed to measure PWV. Results from in-vivo invasive studies on animals (N = 2, male porcine) and non-invasive studies on human participants (N = 60, healthy, age: 20 – 56 years) indicate a lower coefficient of variation (~7% (from ~8%) at diastole, and ~10% (from ~30%) at systole) in the beat-to-beat estimates of PWV after eliminating PWR. Additionally, the mean values obtained from reflection-free PWV and from theoretical estimates of PWV were statistically assessed, and the results indicated no significant difference between the means (p > 0.05) at both diastole and systole. In comparison, the PWV (with reflections) revealed a statistically significant difference between the means (p < 0.05). The method and system proposed in this thesis would potentially strengthen the on-going global efforts of achieving reliable measurements of EVA markers for CVD risk stratifications.