Heart Sound Segmentation using Envelogram in Hadamard Domain

A new framework for heart sound segmentation is proposed. The segmentation algorithm is based on the envelope created using the energy of the first component of the transformed signal, which intensifies the first and second heart sound while attenuating the effects of noise and murmur. Based on the created envelogram, a threshold is calculated to eliminate the effects of noise and low-intensity signals. The estimation of cardiac cycle length using statistical characteristics of the envelope signal is developed. Finally, individual cardiac cycles are extracted from the heart sound. The proposed method showed promising results and high robustness across a wide range of heart sounds.