Automatic Detection Of Heart Sounds And Clicks With Murmurs

This paper proposes an automatic heart sound and click detection algorithm. The proposed approach can accurately detect the first heart sound (S-1), the second heart sound (S-2), and systolic or any diastolic click that occurs. Phonocardiogram data are divided into short data segments in several milliseconds where defined features carrying essential information regarding loudness and marked time locations are extracted. These features are examined through an expert system to identify S-1, S-2, systoles, diastoles, and clicks. The performance of the algorithm was validated with a variety of clinical heart sounds episodes featuring murmurs and clicks.