The Heart Defect Analysis Based on PCG Signals Using Pattern Recognition Techniques

The graphical recording of the heart sounds and murmurs is called Phonocardiogram or PCG and the machine is so called phonocardiograph. It has an important role in the proper and accurate diagnosis of the heart defects. It requires highly and experienced professionals to read the phonocardiogram, as usually with the stethoscope. The paper is about the implementation of a diagnostic system as a detector and classifier; for heart diseases. Various heart sound samples are classified using Support Vector Machine (SVM), K Nearest Neighbour (KNN), Bayesian and Gaussian Mixture Model (KNN) Classifiers. The output of the system is the classification of the sound as either normal or abnormal and if it is abnormal, what type of abnormality is present. In the proposed method, time domain and frequency domain features are extracted. Various frequency domain features such as energy, mean, variance and Mel Frequency Cepstral Coefficients (MFCC) are analysed.