An effective method to identify various factors for denoising wrist pulse signal using wavelet denoising algorithm

Area of Science:

• Biomedical Signal Processing
• Wavelet Theory
• Data Analysis

Background:

• 웨이브렛 파워 스펙트럼 (WPS) is a non-invasive technique for human health assessment.
• Signal acquisition in WPS often includes noise, necessitating signal cleaning for accurate analysis.
• High signal-to-noise ratio in WPS is crucial for reliable disease diagnosis.

Purpose of the Study:

• To establish an effective approach for selecting optimal factors in wavelet denoising algorithms for WPS.
• To enhance the signal-to-noise ratio of WPS signals through appropriate wavelet denoising.
• To clarify the specific wavelet and factor selections for WPS applications.

Main Methods:

• Systematic variation of all wavelet denoising factors.
• Evaluation of denoising performance using Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Percentage Root Mean Square Difference (PRD), and Fit Coefficient.
• Utilizing the Universal Wavelet Transform (UWT) for denoising.

Main Results:

• The study successfully identified optimal parameters for wavelet denoising of WPS.
• Selection of the db9 mother wavelet, the 'sure' threshold function, and a single rescaling function demonstrated superior performance.
• Computerized WPS analysis confirmed the effectiveness of the proposed parameter selection approach.

Conclusions:

• The developed methodology is effective for denoising WPS signals across various morphological patterns.
• Optimized wavelet denoising significantly improves WPS data quality for diagnostic applications.
• This approach provides a robust framework for enhancing biomedical signal processing techniques.