Surface versus untargeted intramuscular EMG based classification of simultaneous and dynamically changing movements

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Pattern recognition myoelectric control is advancing for clinical use.
• Current systems primarily tested on sequential, steady-state data.
• Limited understanding of performance with simultaneous, dynamic movements.

Purpose of the Study:

• Investigate pattern recognition for simultaneous, dynamic movements.
• Compare surface and intramuscular EMG signal performance.
• Assess EMG signal robustness to data modulation.

Main Methods:

• Ten able-bodied subjects performed dynamic forearm contractions.
• Recorded surface and intramuscular EMG signals concurrently.
• Tested performance on nonmodulated and modulated contraction profiles.

Main Results:

• Achieved up to 93% accuracy on nonmodulated tasks.
• Surface and intramuscular EMG showed similar performance on nonmodulated data.
• Intramuscular EMG performance decreased with modulated data compared to surface EMG.

Conclusions:

• Pattern recognition can resolve simultaneous, dynamic movements.
• Surface and intramuscular EMG offer comparable performance for nonmodulated data.
• Intramuscular EMG shows potential for implantable electrodes, requiring careful training data selection.