High performance heart sound segmentation algorithm based on Matching Pursuit

The segmentation of heart sound signals is the first step of analysis for many applications such as computer-aided auscultation, heart sound classification or signal compression. In this paper we present a novel heart sound segmentation technique based on the Matching Pursuit algorithm. We also show that Gabor atoms can provide highly effective representation of audio signals. The performance of the proposed method has been evaluated using a dataset composed of 151 sound attacks comprising several cardiac pathologies. The reported algorithm showed high performance, achieving a detection rate of 97.5% and 96% for heart sound onsets and offsets respectively.