An Improved Wavelet Threshold Shrinkage Algorithm for Noise Reduction of Heart Sounds

Abstract

To compensate for the deficiencies of traditional wavelet threshold shrinkage denoising algorithm and obtain the heart sounds with high SNR, an improved wavelet threshold shrinkage algorithm is proposed in this paper. The algorithm uses a new subband adaptive threshold and a new threshold function to quantify the detail coefficients of wavelet decomposition of noisy heart sounds, and its denoising effect is assessed with signal to noise ratio (SNR) and the root mean square error (RMSE). First, the proposed subband adaptive threshold and the several traditional thresholds are applied to the noise reduction of soft threshold function respectively to verify the superiority of the proposed subband adaptive threshold by the comparison of the denoising effects of 22cases of standard heart sounds in 3M database. Second, the new subband adaptive threshold is applied to the noise reduction of the new threshold function based on genetic algorithm optimization and traditional soft and hard threshold functions respectively to validate the denoising performance of the proposed threshold function. Finally, the improved threshold shrinkage denoising algorithm is used to do noise reduction for 20 cases of the clinical heart sounds. The results indicate that the improved wavelet threshold shrinkage denoising algorithm can eliminate noise more effectively, and has strong clinical value.