Topological data analysis improves estimation of muscle fatigue and contraction level from surface electromyography...

Area of Science:

• Biomedical Engineering
• Data Science
• Neuroscience

Background:

• Muscle fatigue assessment is challenging due to the interplay between fatigue and contraction intensity.
• Current non-invasive methods like surface electromyography (sEMG) struggle to reliably differentiate these factors.
• There is a need for advanced analytical techniques to improve the interpretation of sEMG data for muscle function evaluation.

Purpose of the Study:

• To investigate the efficacy of topological data analysis (TDA) in conjunction with sEMG for assessing muscle function.
• To compare TDA's performance against traditional Fourier transform analysis in estimating muscle fatigue and contraction levels.
• To explore TDA's potential for real-time, non-invasive muscle function monitoring.

Main Methods:

• Recorded sEMG data from the first dorsal interosseus muscle during sustained and discrete contraction experiments.
• Applied TDA and Fourier transform analysis to the sEMG time series data.
• Utilized deep network regression and classification models to analyze TDA-derived features and traditional sEMG features.

Main Results:

• TDA measures demonstrated strong correlations with contraction time in sustained contractions (R² up to 0.95).
• TDA-enhanced deep network regression improved muscle contraction time estimation (R² = 0.83).
• TDA-based classification achieved 71.0% accuracy in distinguishing discrete contraction levels, significantly outperforming traditional methods.

Conclusions:

• Topological data analysis presents a robust method for estimating muscle contraction level and fatigue from sEMG signals.
• TDA offers advantages over traditional frequency and amplitude analysis for sEMG data interpretation.
• Further research is warranted to facilitate the clinical implementation of TDA for muscle function assessment.