Abstract
The paper presents the results of a signal processing approach to detect and isolate systolic murmurs. The identification of the first and second heart sounds and separating systole and diastole from a complete cardiac cycle were successfully carried out through wavelet analysis using an orthogonal Daubechies (db6) wavelet as the mother wavelet. At the fifth level of decomposition, S1 and S2 were effectively separated from systolic heart murmurs. A quantitative measure of signal energy was developed and used to determine the boundaries of S1 and S2 sounds and to isolate systolic murmurs. The energy index can be also used to delineate the intensity and configuration of a systolic murmur. We have examined and reported in this paper the performance of this approach by examining few known clinical systolic murmurs: atrial septal defect (ASD), ventricular septal defect (VSD), and mitral valve prolapse (MVP).