Performance of Hammerstein Spline Adaptive Filtering Based on Fair Cost Function for Denoising Electrocardiogram...

Area of Science:

• Biomedical Signal Processing
• Adaptive Filtering
• Nonlinear Systems

Background:

• Linear filters have limitations in real-world applications, especially for complex biological signals like ECG.
• Adaptive nonlinear filtering offers a promising approach to address individual variations and noise in biomimetic systems.
• ECG signal denoising is crucial for accurate cardiac diagnosis and monitoring.

Purpose of the Study:

• To propose a simplified adaptive nonlinear filtering approach for denoising electrocardiogram (ECG) signals.
• To develop an individual biomedical filter capable of adapting to unique ECG characteristics and removing noise effectively.
• To enhance the convergence and performance of adaptive filters using a Fair cost function.

Main Methods:

• A Hammerstein spline adaptive filter (HSAF) architecture combining nonlinear and linear blocks was designed.
• The Fair cost function was integrated to improve convergence smoothness and efficiency.
• The affine projection algorithm (APA) with the Fair cost function was employed for denoising contaminated ECG signals.
• The MIT-BIH 12-lead database was used, with noise modeled by Cauchy distribution.

Main Results:

• The proposed HSAF-APA-Fair algorithm demonstrated reduced estimation error in denoising ECG signals.
• The algorithm showed improved performance compared to conventional least mean square (LMS)-based algorithms.
• The Fair cost function contributed to faster convergence and enhanced denoising accuracy.

Conclusions:

• The developed adaptive nonlinear filter provides an effective method for individual ECG signal denoising.
• The HSAF-APA-Fair algorithm offers a robust solution for removing noise while preserving essential ECG characteristics.
• This approach advances the field of biomedical signal processing for personalized healthcare applications.