Principal component analysis (PCA) approach to segment primary components from pathological phonocardiogram

Heart auscultation (interpretation of heart sounds) is the primary tool used in screening patients for heart pathology, and they are usually found in the primary health care. In this paper, a method based on principal component analysis is proposed for segmenting heart sounds. Firstly, the signal is filtered to remove low frequency noises and decimated to consider only the frequencies which are of clinical significance. Then principal component analysis is used to extract the feature set which is envelope extracted using Shannon energy and sub-divided into individual cardiac cycles using variance based algorithm. Finally, the envelope is segmented by using cardiac periods of the signal. Any false segmentation is eliminated according to the subjective knowledge of the heart sounds. Experimental results show that the proposed statistical approach performs well for both normal and pathological heart sounds with segmentation accuracy of 97.7%.