High-yield decomposition of surface EMG signals

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Surface electromyography (sEMG) is crucial for understanding muscle activity.
• Accurate decomposition of sEMG into motor unit action potential trains (MUAPTs) is challenging.
• Existing methods often lack precision, especially in complex physiological conditions.

Purpose of the Study:

• To develop and validate an enhanced artificial intelligence (AI) algorithm for automatic sEMG decomposition.
• To assess the reliability and accuracy of this AI technology in decomposing MUAPTs.
• To evaluate the technology's performance across various muscles and force levels, including challenging conditions like adipose tissue.

Main Methods:

• Utilized a novel, enhanced AI algorithm processing four-channel sEMG signals from a compact sensor.
• Tested the algorithm on isometric contractions from five muscles up to 100% maximal force, including subjects with over 1.5cm adipose tissue.
• Employed a new decomposition-reconstruction accuracy measurement method, validated by the two-source method.

Main Results:

• The AI algorithm successfully decomposed 20-30 MUAPTs, occasionally up to 40, across different muscles and force levels.
• Achieved an average decomposition accuracy of 92.5%, reaching up to 97% for MUAPT firings.
• Demonstrated reliable performance even with significant adipose tissue coverage.

Conclusions:

• The developed technology reliably performs high-yield decomposition of sEMG signals during isometric contractions up to maximal force.
• The system's small sensor size, high yield, and accuracy make it suitable for motor control studies and clinical applications.
• This advancement offers a promising tool for detailed analysis of neuromuscular function.