Identification and features extraction of systolic and diastolic murmurs

This paper presents an automatic heart murmur detection and delineation algorithm. Multiple cardiac cycles with murmurs are processed in two steps. In the first step, the first and second heart sounds are detected and identified to label cardiac cycles as the first heart sound (S₁), followed by a systole period, the second heart sound (S₂), a diastole period, and then repeated with S₁ and the rest. In the second step, potential murmur occurs within systole or diastole are identified and delineated with generated features such as timing, duration, shape configuration and pitch frequency, which are useful measures to assist clinical diagnosis. The performance of the proposed algorithm was examined and confirmed with six different types of diastolic and systolic murmurs of clinical data.