Murmur Detection in PCG signals using DWT Entropy and Feature Clustering

Phonocardiogram (PCG) signals, their feature extraction and segmentation to diagnose several heart disorders and developing detection models is the primary area of research. Multiple challenges are associated with the recording and preprocessing of these data due to inherent environmental noise and subject wise variation in heart sound. In this work two different approaches based on Fast Fourier Transform (FFT) and Discrete Wavelet Transform (DWT) is proposed and algorithms are developed to classify the normal subjects and the subjects having PCG signal with high level of murmur due to heart disorders. The results are finally processed through cluster-based analysis of the DWT components entropy value to discriminate the normal subjects and murmur inhibited PCG data.