Deep Learning Based Classification of Heart Diseases from Heart Sounds

Abstract

Heart problems have become one of the prevalent impacts on human health. Heart sound identification techniques play an essential role in predicting heart attacks as a non-invasive assistant diagnostic tool. The phonocardiogram (PCG) signal, a digital recording of heart sounds, was analyzed in this paper and classified into four classes: Normal signal, Murmur signal, Artifact signal, and Extra systole signal. The following are the primary considerations: the relationship between heart sounds and cardiovascular diseases: Denoising, segmentation, feature extraction, and classification are some of the core technologies used in the processing and analysis of heart sound signals, with a focus on the applications of deep learning algorithms in heart sound processing. Mel-frequency Cepstrum (MFCC) was used to extract different features for classifying the dataset. The most conspicuous characteristics are chosen and employed in training the deep learning algorithm for the automatic classification of heart diseases. The deep learning methodologies employed in the study were CNN and LSTM. The findings of the experiment reveal that the efficiency of the procedure appropriate diagnosis of cardiac illnesses using the heart sound dataset is excellent in accuracy, making it appropriate for real-time applications.