Time-frequency analysis of heart sounds

From a study of the physical characteristics of heart sounds and human hearing capabilities, it is seen that only the highly trained ear may be able to detect heart sounds events. Also from a study of heart sound signals nature and the frequency domain analysis based on direct application of the Fourier transform technique, it is seen that this technique is inappropriate because important time events and features such as frequency change will be lost. To solve the above mentioned problems the wavelet transform, which is a new time-scale technique suitable for transient signal detection is applied to phonocardiogram signals taken from group of patients. It has been shown that using this new technique the separation between the normal and abnormal subject could easily be obtained. Moreover, a separation between the high frequency systolic and diastolic murmurs, and the low frequency systolic and diastolic murmurs can be achieved. Furthermore, the temporal location of the auscultatory gap can be detected, which is usually very difficult to discern even to the trained ear