A Study on the Segmentation of Heart Sound

In order to perform the automatic analysis of heart sounds, segmentation of first (sl) and second (s2) heart sounds is a very important task. In this study, an effective segmentation method of the heart sound signals is proposed based on empirical mode decomposition and Hilbert transform. Initially, the signal is filtered by using empirical mode decomposition to remove the ambient and physiological noise interference introduced during heart sound acquisition. Next, the envelope of the heart sound is extracted by Hilbert transform. Then, the timing information of sl and s2 are identified based on biomedical domain knowledge. The method has been tested with data collected from normal and coronary artery disease Subjects. Experimental results show that 92.63% and 86.67% segmentation accuracies are obtained for normal and coronary artery disease Subjects, respectively.