Selecting the optimal movement subset with different pattern recognition based EMG control algorithms.

Area of Science:

• Biomedical Engineering
• Rehabilitation Engineering
• Human-Computer Interaction

Background:

• Pattern Recognition (PR)-based Electromyography (EMG) controllers enhance multi-functional upper-limb prostheses with intuitive control and fast reaction times.
• Current PR systems lack optimized training and deployment protocols tailored to individual amputees' needs.
• Subject-specific training requires excluding poorly performing movements and focusing on achievable ones.

Purpose of the Study:

• To propose a method for selecting the optimal subset of movements for individual amputees using PR-based EMG controllers.
• To identify movements with the worst performance for targeted training and improved prosthetic control.
• To investigate the impact of different feature extraction and classification methods on movement subset selection.

Main Methods:

• Collected EMG signals from four transradial amputees.
• Utilized various feature extraction and classification techniques to analyze EMG data.
• Developed and applied a novel approach for selecting the best subset of controllable movements.

Main Results:

• Achieved an average accuracy greater than 97.5% across all subjects for selecting the best subset of movements.
• Demonstrated the effectiveness of the proposed method in identifying subject-specific movement capabilities.
• Showcased that the choice of features/classifiers can influence the optimal movement subset selection.

Conclusions:

• The proposed method effectively identifies optimal movement subsets for PR-based EMG prosthetic control, enhancing personalized training.
• Targeted identification of challenging movements allows for focused rehabilitation, improving overall prosthetic functionality.
• The study highlights the importance of considering feature/classifier choices for optimizing prosthetic control systems.