Automatic extraction of physiological features from vibro-acoustic heart signals: correlation with echo-doppler

Abstract

The mechanical processes within the cardiovascular system produce low-frequency vibrations and sounds. These vibro-acoustic signals carry valuable physiological information that can be potentially used for cardiac monitoring. In this work, heart sounds, apical pulse, and arterial pulse signals were simultaneously acquired, along with electrocardiogram and echo-Doppler audio signals. Processing algorithms were developed to extract temporal and morphological feature from the signals. Spectral analysis was used to reconstruct the Doppler sonograms and estimate reference values. A good agreement was observed between systolic and diastolic time intervals estimated by both methods. Strong beat-to-beat correlations were shown both in rest and during pharmacological stress test. The results demonstrate the technological and medical feasibility of using automatic analysis of vibro-acoustic heart signals for continuous non-invasive monitoring of cardiac functionality