Low Cost Portable Diagnostic Device for Automatic Classification of the Abnormal Cardiac Sound using PGC Recording

Abstract

Cardiovascular diseases are becoming one of the most common causes of mortality all over the world these days. Medical experts and professionals use stethoscope for proper analysis of the cardiac sound. This conventional method required a lot of and also involve medical experts. This paper presents a low cost, portable device which can automatically classified the heart condition by recording the heart sound of the patient. The device is developed in such a way that a non-medical person can use it for the purpose of initial screening of the heart condition of the patients. The proposed device is based on supervised classifier which help in identifying a recorded heart sound either as normal or abnormal heart sound. Supervised classification model is developed on the basis of discriminatory features that are extracted using cepstrum analysis of the heart sound. The proposed method has achieved an accuracy of 97% in correctly classifying a heart sound PCG signal as normal and abnormal. This make the developed device to use in dispensary for the initial screening of the patients.