An Experimental Method for Bio-Signal Denoising Using Unconventional Sensors

Area of Science:

• Biomedical Engineering
• Signal Processing

Background:

• Current bio-signal denoising methods often rely on simulated environments, demanding high computational power and potentially distorting signals.
• Existing techniques require prior knowledge of noise characteristics or signal periodicity, complicating noise estimation.

Purpose of the Study:

• To develop and evaluate an experimental method for efficient, real-time bio-signal denoising.
• To address limitations of simulated environments and improve noise estimation in bio-signal processing.

Main Methods:

• Implementation of unconventional electric field sensors to generate a noise replica.
• Utilizing the noise replica to derive the ideal Wiener filter transfer function for noise reduction.
• Experimental validation using human bio-signals (ECG, EMG, EOG) and the MIT-MIH arrhythmia database.

Main Results:

• The proposed method achieved significant noise attenuation: 26.4 dB for ECG, 21.2 dB for EMG, and 40.8 dB for EOG.
• Performance was evaluated using power spectral density, signal-to-noise ratio, and mean square error.
• Comparison with state-of-the-art methods demonstrated the efficacy of the combined approach.

Conclusions:

• The developed combined approach is suitable for real-time noise reduction in various bio-signals.
• This experimental method offers an effective alternative to simulation-based denoising techniques.
• The use of electric field sensors provides a practical solution for accurate noise replica generation.