Investigation of some features for preliminary detection of coronary artery disease using electronic stethoscope

Recent research shows that the diastolic part of cardiac cycle contains very weak murmurs caused by turbulent flow in the coronary artery diseases. This paper deals with the use of electronic stethoscope for the preliminary detection of coronary artery diseases (CAD). 3M Littmann 3200 model is used for recording of cardiac sounds of CAD and non CAD category. Cardiac sounds were analyzed in both time and frequency domain. Six features including two time domain and four frequency domain were extracted from each sound. Out of these features two frequency domain features namely spectral centroid and spectral roll off were identified as the potential features for discriminating CAD and non CAD sounds.