Evaluation of Performance Metrics and Denoising of PCG Signal using Wavelet Based Decomposition

Phonocardiogram (PCG) signal contains significant bio-acoustic information reflecting the operation of the heart, and is used to detect the various diseases related to heart valve. But it is highly susceptible to noise, and the sources of noise includes lung and breath sounds, noise from contact between the recording device and skin, environmental noise, etc. Hence, denoising of PCG signal is very important for the proper diagnosis of heart diseases. In this paper, a discrete wavelet transform (DWT) based threshold tuning is investigated to deliver denoised PCG signal. The performance of the denoising algorithm is evaluated using different metrics such as mean-square error, normalized-mean-square error, root-mean-square error, percentage root-mean-square difference, and signal-to-noise ratio by determining the most suitable parameters (wavelet family, level of decomposition, and thresholding type) for the denoising process. The evaluation results obtained from the different metrics gives the best denoising performance from the reconstructed PCG signal.