Classification of normal and abnormal heart sounds for automatic diagnosis

Body auscultation is an easy and non-invasive method for detection of diseases in human body. The conventional method is a bit time consuming and requires professionals for diagnosis. Automatic diagnosis of diseases using heart sound can be of great help in the rural areas where professional help is not available. The proposed work presents an automatic and efficient method of diagnosis and classification using heart sound. Mel Frequency Cepstral Coefficient (MFCC) features are extracted from heart sounds for diagnosis. Supervised classification method is used to separate the normal and abnormal heart sound for detection of diseases. The proposed method was tested on a comprehensive database of heart sounds and achieved accuracy of 97.50% during classification process. The experiment results indicates that the proposed method is efficient for classification of healthy/unhealthy heart sounds and computationally cheap making it suitable for real time applications.