Individual finger classification from surface EMG: Influence of electrode set

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Decoding individual finger movements is crucial for advanced prosthetics and human-computer interfaces.
• Surface electromyography (sEMG) offers a non-invasive method for muscle activity detection.
• Optimizing sEMG electrode placement is essential for efficient and accurate signal acquisition.

Purpose of the Study:

• To minimize the number of sEMG channels required for accurate decoding of individual finger isometric contractions.
• To determine optimal electrode locations and recording configurations.
• To evaluate different electrode configurations for classifying isometric flexion and extension movements.

Main Methods:

• Nine healthy subjects performed cyclical isometric contractions of individual fingers.
• Surface electromyography (sEMG) signals were recorded from forearm muscles using a 192-channel matrix.
• Classification accuracy was assessed using linear discriminant analysis (LDA) with four electrode configurations: single array, dual arrays, barycenter-based, and full channel set.

Main Results:

• Classification accuracy varied significantly across different electrode configurations (F=14.67, p<0.001).
• The barycenter approach and dual linear arrays of 8 electrodes yielded the highest classification accuracies, exceeding 82% success rate.
• These optimized configurations performed comparably to using all recorded channels.

Conclusions:

• Dual linear arrays of 8 electrodes represent an optimal configuration for classifying individual finger isometric flexion and extension.
• This configuration balances high classification accuracy with reduced computational time and simplified electrode positioning.
• The findings contribute to developing more efficient and practical sEMG-based control systems.