Publication

Abstract : Healthcare industry is growing day by day giving support to patients both in a rural area and urban area. Around 17.9 million death was reported due to heart diseases in past years. Development of digital phonocardiogram as a primary diagnostic device has a big future in the present scenario of the world. Most of the heart valve defect can be identified using this technique. But the heart sound contains different background noises. In this paper, a robust method of using Convolution Neural Network (CNN) is to classify the heart sound to normal and abnormal signals. The background noises were removed use smoothing filters and wavelet denoising technique as our primary work. The obtained accuracy while using this methodology was about 97%. This is very helpful to physicians for diagnosis at the initial stage.