Dual Window Pattern Recognition Classifier for Improved Partial-Hand Prosthesis Control.

Area of Science:

• Biomedical Engineering
• Rehabilitation Engineering
• Human-Computer Interaction

Background:

• Controlling myoelectric partial-hand prostheses is challenging, especially when preserving wrist motion.
• Electromyogram (EMG) signals from wrist movement can interfere with finger control signals.

Purpose of the Study:

• To develop a training protocol and classifier that accommodates wrist motion for improved myoelectric partial-hand prosthesis control.
• To minimize system latency and maximize classification accuracy.

Main Methods:

• A dual-window EMG analysis (long and short) was developed and tested.
• Seventeen non-amputee and two partial-hand amputee subjects participated.
• EMG data included static/dynamic wrist motion and intrinsic/extrinsic hand muscle signals.
• Real-time classification techniques were evaluated using virtual tasks and ANOVA.

Main Results:

• Including wrist motion and intrinsic hand muscle EMG reduced pattern recognition classification error by 35%.
• Classification delay or majority voting techniques improved real-time task completion and selection efficiency.
• The dual-window classifier reduced time and attempts for grasp selections in various wrist positions.
• Amputee subjects showed improved task timeout rates and fewer grasp selection attempts.

Conclusions:

• The proposed dual-window classifier and training protocol show promise for enhancing partial-hand prosthesis control.
• These techniques can help restore function more effectively for individuals with partial-hand amputations.
• Integrating wrist motion signals improves the usability and performance of myoelectric prostheses.