Adaptive filtering for ECG rejection from surface EMG recordings

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Surface electromyograms (SEMG) are frequently contaminated by electrocardiogram (ECG) signals.
• This ECG noise obscures the spectral content of SEMG, hindering accurate analysis of muscle fatigue.
• Existing noise reduction methods often fail to account for dynamic changes in ECG characteristics.

Purpose of the Study:

• To develop and validate an adaptive filtering algorithm for robust ECG noise removal from SEMG.
• To optimize ECG electrode placement for improved noise signal similarity to SEMG artifacts.
• To evaluate the algorithm's effectiveness in analyzing muscle fatigue parameters.

Main Methods:

• An adaptive filtering algorithm, specifically a simplified Fast Recursive Least Square (FRLS), was developed.
• ECG electrodes were positioned to match the artifact shape in SEMG recordings.
• The algorithm's performance was tested on 28 erector spinae SEMG recordings and applied to 16 fatigue test recordings.

Main Results:

• The simplified FRLS adaptive filtering algorithm demonstrated superior ECG rejection from SEMG.
• Denoised SEMG signals showed significantly different initial energy and mean power frequency (MPF) values compared to raw signals.
• Fatigue analysis on denoised SEMG revealed clearer trends in energy increase and MPF decrease over time.

Conclusions:

• The proposed adaptive filtering method effectively removes ECG artifacts from SEMG signals.
• Denoising enhances the accurate assessment of muscle fatigue parameters, such as energy and MPF evolution.
• This technique improves the reliability of electrophysiological signal analysis in the presence of ECG noise.