An Automatic Wavelet Selection Scheme for Heart Sounds Denoising

The phonocardiograms (PCGs), recording of heart sounds, have many advantages over traditional auscultation in that they may be replayed and analyzed for spectral and frequency information. PCG is not a widely used diagnostic tool as it could be. One of the major problems with PCG is noise corruption. Many sources of noise may pollute a PCG signal including lung and breath sounds, environmental noise and blood flow noises which are known as murmurs. Murmurs contain many information on heart hemodynamic which can be used particularly in detecting heart valve diseases. Therefore such diseases can be automatically diagnosed using Murmurs. However, the first step before developing any automated system using Murmurs is the denoising and the segmentation of the PCG signal from which murmurs can be separated. Different algorithms have been developed in the literature for denoising and segmenting the PCG signal. A robust segmentation algorithm must have a robust denoising technique. The wavelet transform (WT) is among the ones which exhibits very high satisfactory results in such situations. However, the selection of level of decomposition and the mother wavelet are the major challenges. This paper proposes a novel approach for automatic wavelet selection for heart sounds denoising. The obtained results on real PCG signal embedded in different white noise intensity showed that the proposed approach can successfully and consistently extract the main PCG sound components (sound component S1 and sound component S2) from various types of murmurs with good precision.