Self-correcting pattern recognition system of surface EMG signals for upper limb prosthesis control.
IEEE Transactions on Bio-Medical Engineering
|March 25, 2014
Summary
This study introduces a new algorithm to improve prosthetic hand control by reducing motion misclassifications. The enhanced pattern recognition system boosts accuracy in real-world conditions for upper limb prosthesis users.
Area of Science:
- Biomedical Engineering
- Neuroscience
- Rehabilitation Technology
Background:
- Surface electromyography (sEMG) pattern recognition for upper limb prosthetics lacks real-world robustness.
- Current methods struggle with variations in daily use, limiting clinical viability.
Purpose of the Study:
- To develop and evaluate a novel postprocessing algorithm for detecting and removing misclassifications in sEMG-based motion intent recognition.
- To enhance the clinical viability and robustness of upper limb prosthesis control systems.
Main Methods:
- An artificial neural network was trained using maximum likelihood and mean global muscle activity to identify erroneous classification decisions.
- The proposed algorithm was compared against four existing postprocessing methods.
- Experiments included able-bodied and amputee subjects under various nonstationary conditions (e.g., varying contractions, electrode shifts, user variability).
Main Results:
- The novel postprocessing system significantly improved classification accuracy, with improvements ranging from 4.8% to 31.6% across different scenarios.
- The system effectively reduced misclassifications into incorrect active classes.
- Performance gains were observed even with challenges mimicking real-life prosthesis use.
Conclusions:
- The developed postprocessing algorithm demonstrates a promising approach to enhance the robustness of hand prosthesis control.
- By reducing misclassifications, the system offers improved controllability for upper limb prosthetics in real-world applications.
