Analysing heart murmurs using time-frequency methods

Abstract

Discusses the use of time-frequency methods in the detection and analysis of heart murmurs in Phono-CardioGraphic (PCG) signals. These heart sounds can yield important diagnostic information and are measured in a non-invasive fashion. The methods employed are developed heuristically based on a model of heart sounds, but their connection to existing techniques is also presented. The method exploits averaged versions of the Pseudo-Wigner-Ville distribution (PWVD). The algorithms are shown to detect two types of heart murmur and to be able to distinguish between them, a task which requires an experienced human listener.