Cardiac Auscultation Classification

With increase in the number of people getting affected from cardiovascular diseases nowadays, it is necessary to diagnose the disease in the earlier stages before it gets severe, in order to prevent serious problems in the future which might be life threatening. Early detection can be done through computer aided analysis using the data obtained from Phonocardiogram (PCG) auscultation obtained from a stethoscope. This paper takes five PCG signals - normal, artifact, murmur, extra systole and extrahls that are obtained as datasets from Peter J Bently Challenge. They are analysed by Machine Learning (ML) and Deep Learning (DL) algorithms: SVM, RNN-LSTM, CNN while their performance is compared and studied.