Wavelet analysis of phonocardiograms: differences between normal and abnormal heart sounds

Abstract

Phonocardiography-based diagnosis can be improved by modern techniques of signal processing. This work investigates the analysis of phonocardiograms using the time-frequency distribution generated by the cochlear wavelet transform. We show here a comparative study of normal and abnormal heart sounds, whose scalograms exhibit noticeable morphological differences in terms of duration and spectral composition of sounds. These differences suggest that the cochlear-wavelet-based scalogram can become an interesting approach to design automatic diagnosis systems.