Heart sound feature extraction based on wavelet singular entropy

Abstract

After analyzing the advantages and disadvantages of current methods of heart sound feature extraction, a new method based on wavelet transform, singular value decomposition and information entropy is put forward. In this method, firstly the heart sound is decomposed by wavelet transformation. Then the singular value of the sub-bands containing heart sound information is obtained by decomposition. Finally, according to the constructed wavelet singular entropy, the entropy of the above singular value is obtained. By comparing the wavelet singular entropy of normal heart sound signal with the several heart sound signals with pathological information, wavelet singular entropy can be found a good characterization of heart sound. © 2014 Springer-Verlag.