Prediction of Dexterous Finger Forces With Forearm Rotation Using Motoneuron Discharges.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Engineering

Background:

• Electromyogram (EMG) decomposition decodes finger movement for neural-machine interfaces (NMIs).
• Forearm rotation alters motor unit action potential (MUAP) shape, degrading EMG decomposition and motion decoding accuracy in real-time.
• Accurate motor unit (MU) discharge information is crucial for dexterous multi-finger force prediction.

Purpose of the Study:

• To develop an adaptive method for accurate MU discharge information extraction across forearm rotation positions under real-time conditions.
• To enhance dexterous multi-finger force prediction for NMI applications.

Main Methods:

• Utilized FastICA-based EMG decomposition.
• Developed a method to obtain multiple separation vectors per MU at different forearm positions during initialization.
• Implemented adaptive real-time extraction of MU discharge information using the nearest forearm position's separation vector.

Main Results:

• The proposed adaptive method significantly outperformed previous methods using a single constant separation vector or conventional EMG amplitude information.
• Demonstrated significantly higher coefficient of determination (R²) and lower root mean squared error (RMSE) in force prediction.
• Validated the feasibility and effectiveness of the adaptive method for real-time MU discharge extraction during forearm rotation.

Conclusions:

• The developed adaptive method effectively extracts MU discharge information during forearm rotation, improving real-time force prediction for NMIs.
• This approach enhances the robustness of EMG decomposition against forearm position variations.
• Further development could significantly advance continuous dexterous motion decoding in realistic NMI applications.