• 1. School of Automation Science and Engineering, South China University of Technology, Guangzhou 510641, P. R. China;
  • 2. Guangdong Engineering Research Center of Cloud-Edge-End Collaboration Technology for Smart City, Guangzhou 510641, P. R. China;
  • 3. School of Electronics and Information Engineering, South China University of Technology, Guangzhou 510641, P. R. China;
  • 4. Guangdong Second Provincial General Hospital, Guangzhou 510317, P. R. China;
MO Hongqiang, Email: hqiangmo@scut.edu.cn
Export PDF Favorites Scan Get Citation

Adaptive filtering methods based on least-mean-square (LMS) error criterion have been commonly used in auscultation to reduce ambient noise. For non-Gaussian signals containing pulse components, such methods are prone to weights misalignment. Unlike the commonly used variable step-size methods, this paper introduced linear preprocessing to address this issue. The role of linear preprocessing in improving the denoising performance of the normalized least-mean-square (NLMS) adaptive filtering algorithm was analyzed. It was shown that, the steady-state mean square weight deviation of the NLMS adaptive filter was proportional to the variance of the body sounds and inversely proportional to the variance of the ambient noise signals in the secondary channel. Preprocessing with properly set parameters could suppress the spikes of body sounds, and decrease the variance and the power spectral density of the body sounds, without significantly reducing or even with increasing the variance and the power spectral density of the ambient noise signals in the secondary channel. As a result, the preprocessing could reduce weights misalignment, and correspondingly, significantly improve the performance of ambient-noise reduction. Finally, a case of heart-sound auscultation was given to demonstrate how to design the preprocessing and how the preprocessing improved the ambient-noise reduction performance. The results can guide the design of adaptive denoising algorithms for body sound auscultation.

Citation: MO Hongqiang, TIAN Xiang, LI Bin, TIAN Junzhang. Improving adaptive noise reduction performance of body sound auscultation through linear preprocessing. Journal of Biomedical Engineering, 2024, 41(5): 969-976. doi: 10.7507/1001-5515.202307058 Copy

  • Previous Article

    Enhancement algorithm for surface electromyographic-based gesture recognition based on real-time fusion of muscle fatigue features
  • Next Article

    Heart sound classification algorithm based on bispectral feature extraction and convolutional neural networks