Comparison of six envelope extraction methods based on abnormal heart sounds

摘要

How to extract envelopes accurately is the most important problem in analyzing heart sound signals, especially for abnormal heart sound signals. An effective envelope extraction is a key to the detection of S1 and S1 periods and noise separation, moreover, for estimating the type of heart disease. This paper compares six efficient envelope extraction methods based on abnormal heart sounds, with comparisons after the envelope endpoints auto-detection and the error analysis. The result indicates that the envelope extracted with Wavelet-Hilbert-Huang transform method is the most accurate one with 97% of endpoints detection rate and 16.15 ms error margin, which can provide accurate identification for S1 and S1 durations.