Detecting Abnormal PCG Signals and Extracting Cardiac Information Employing Deep Learning and the Shannon Energy Envelope

Phonocardiogram (PCG) is a computerized system that represents the heart sound recording. PCG reflects the acoustic behavior of the heart graphically through intensity, frequency, time duration, and other valuable information. An acoustic signal like PCG provides supplemental diagnostic information to Electrocardiogram (ECG) by extracting cardiac information from the heart sound those cannot be identified by hearing the heart sound. PCG is an objective and standard evaluation technique that can record the heart sound continuously for a long period of time and can also overcome the human hearing limitation [1] [2]. So, PCG plays a vital role to examine the heart sound as well as cardiac abnormalities which improves the overall diagnosis efficiency. We proposed a PCG classification method using the deep convolutional neural network (CNN). Our proposed method is not only able to classify PCG signals using CNN but can also segment PCG signals using the Shannon energy envelope method. This signal processing technique provides significant information regarding the heart condition that helps to detect heart diseases in the primary phase.