A Robust Envelope Extraction Algorithm for Cardiac Sound Signal Segmentation

This paper presented a new method of envelope extraction algorithms based on wavelet for cardiac sound signal segmentation. In this paper, a new method based on Morlet wavelet is proposed to extract the energy envelope of heart sound. Noise can be decomposed to a different frequency channel for smoothing, thereby reducing the impact of noise. And then physiological criteria of time interval, the ratio of systolic and diastolic duration and phonocardiogram (PCG) collecting position are employed to differentiate S1 and S2. The experiment results show that this envelope extraction algorithm is more sensitive (morlet detection rate of 88.29% versus 66.77% of Hilbert method) in processing signals with low ratio of S/N (signal/noise). This indicates that our method is more robust for heart sound signal detection.