A New Approach on Heart Murmurs Classification with SVM Technique

This paper presents a novel spectral analysis method of heart sounds using the normalized autoregressive power spectral density (NAR-PSD) curve with the support vector machine (SVM) technique for classifying the heart murmurs. The 459 heart sound signals with 196 normal and 263 abnormal sound cases acquired from 6 healthy volunteers and 30 patients were tested. Considering the morphological characteristics of PSD of the heart sound in frequency domain, we proposed two diagnostic parameters fmax and fwidth. In order to detect abnormality of heart sound and to discriminate the heart murmurs, the multi-SVM classifiers composed of four SVM modules were considered and designed. With the proposed cardiac sound spectral analysis method, the high classification performances were achieved by 99.90% specificity and 99.52% sensitivity for classifying normal and abnormal sounds. Furthermore, the rate of discrimination on heart sound murmurs is 86.88% for atrial fibrillation sounds, 89.98% for aortic valvular disorders, and 90% for mitral valvular disorder.