Ranking hand movements for myoelectric pattern recognition considering forearm muscle structure.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Surface electromyography (sEMG) pattern recognition (PR) for hand movements often overlooks individual muscle structures and movement correlations.
• Existing algorithms decode limited hand movements, lacking personalization for optimal performance.

Purpose of the Study:

• To propose a novel method for personally ranking hand movements for sEMG-based PR.
• To optimize the selection of movements based on individual forearm muscle characteristics and electrode placement.

Main Methods:

• Developed a personalized ranking system for hand movements, including finger flexion, extension, and wrist actions.
• Sorted movements based on electrode locations on the proximal and distal forearm.
• Evaluated classification error against the number of desired movements (N_m).

Main Results:

• The maximum number of movements (N_m) with <10% classification error was 20 for proximal forearm placements.
• The maximum N_m with <10% error was 10 for distal forearm placements.
• Personalized ranking significantly impacted decoding accuracy based on individual characteristics.

Conclusions:

• The proposed method effectively identifies an optimized order of hand movements for individual sEMG PR.
• This approach enhances the personalization of prosthetic control and human-machine interfaces.
• Considering individual muscle structures improves the robustness and accuracy of sEMG decoding.