Cross-person decomposition of surface electromyogram for efficient motor unit activity predictions

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Accurate prediction of motor unit (MU) discharge activity from surface electromyogram (sEMG) signals is crucial for understanding neuromuscular control.
• Current MU decomposition methods are person-specific, limiting their generalizability and practical application.

Purpose of the Study:

• To develop and validate a cross-person decomposition framework for sEMG signals.
• To enable generalized and accurate prediction of MU activity across different individuals.

Main Methods:

• Developed a cross-person decomposition framework using synthesized high-density sEMG data.
• Validated the algorithm by comparing predicted MU spike trains with ground truth across multiple metrics.
• Utilized simulated MU firing spike trains convolved with action potential templates from human experimental data.

Main Results:

• Demonstrated strong agreement between predicted and true MU activity with high R2 values (⩾0.95) for population firing rate.
• MU action potential (MUAP) waveform shape and spatial distribution showed high similarity to ground truth.
• Recruitment thresholds exhibited a strong linear relationship (R2 = 0.98 ± 0.006) with minimal error.

Conclusions:

• The developed framework enables efficient cross-person MU decomposition with minimal accuracy loss.
• This research lays the groundwork for generalized, plug-and-play myoelectric systems.
• Findings support applications in neurophysiology, neuroprosthetics, and rehabilitation.