Nonlinear spatio-temporal filter to reduce crosstalk in bipolar electromyogram.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Surface electromyogram (EMG) signals are susceptible to crosstalk, where signals from non-target muscles interfere with recordings.
• Accurate EMG signal decomposition is crucial for numerous applications, including biomechanics, rehabilitation, and prosthetics.
• Existing methods struggle to effectively isolate target muscle activity from crosstalk in bipolar recordings.

Purpose of the Study:

• To develop and validate an innovative nonlinear spatio-temporal filter for estimating target muscle EMG signals.
• To effectively remove crosstalk interference from bipolar EMG recordings using a novel filtering approach.
• To assess the filter's performance across various simulated and experimental conditions.

Main Methods:

• A nonlinear spatio-temporal filter was designed to process signals from two bipolar EMG channels placed over target and crosstalk muscles.
• The filter was trained on calibration data and subsequently applied to new signals.
• Performance was evaluated through simulations (varying tissue properties, electrode placement, force levels) and experiments on human subjects.

Main Results:

• The proposed filter demonstrated a statistically significant reduction in crosstalk, with root mean squared error decreased by approximately 20% in both simulated and experimental data.
• The filter outperformed a conventional blind source separation method in reducing crosstalk.
• Effective crosstalk reduction was observed across all tested conditions.

Conclusions:

• The developed nonlinear filter offers a simple and feasible solution for reducing EMG crosstalk in applications utilizing single bipolar channels.
• This method holds significant potential for improving accuracy in gait analysis, myoelectric fatigue testing, EMG biofeedback rehabilitation, and prosthesis control.