Heart murmurs extraction using the complete Ensemble Empirical Mode Decomposition and the Pearson distance metric

Abstract

Signal processing for pathological heart sound signals can be considered as a fundamental part of the whole process in tele-auscultation systems. In this paper, we employ the CEEMD and the EEMD algorithm to decompose various pathological heart sound signals in the form of phonocardiograph (PCG) signals. Following the decomposition process, we subsequently extract murmurs from the targeted heart sound signals using our proposed technique that based on the Pearson distance metric. Performance analysis of the decomposition algorithms as well as the extraction method is evaluated in terms of delta SNR that signifies variance comparison of targeted signal before and after murmurs extraction. It can be concluded that in general pathological heart sound signals that have been decomposed by the CEEMD algorithm followed by the Pearson distance metric for murmurs extraction, provide the finest murmurs extraction than the EEMD. Additionally, the EEMD algorithm involves smaller number of modes to form the extracted murmurs signal as compared to the CEEMD algorithm. However, employing the CEEMD algorithm produces higher number of shifting procedures causing higher computational complexity than the EEMD algorithm.