Real-time simultaneous myoelectric control by transradial amputees using linear and probability-weighted regression

Area of Science:

• Biomedical Engineering
• Neuroprosthetics
• Rehabilitation Robotics

Background:

• Surface electromyography (EMG) has enabled real-time prosthesis control in transradial amputees, but typically limited to wrist movements.
• Intramuscular EMG shows potential for finer hand and wrist control, yet its efficacy in amputee populations remains under-explored.

Purpose of the Study:

• To evaluate two regression-based control methods using intramuscular EMG in transradial amputees.
• To compare the performance of amputee subjects with able-bodied subjects using these control systems.

Main Methods:

• Two transradial amputees and 16 able-bodied subjects participated.
• Fine-wire intramuscular EMG was recorded from six forearm muscles.
• Subjects controlled three degrees of freedom (DOFs): wrist rotation, wrist flexion/extension, and hand open/close using linear and probability-weighted regression in a Fitts' Law task.

Main Results:

• Initially, amputee subjects showed lower performance metrics than able-bodied subjects.
• One amputee subject demonstrated substantial learning, achieving performance within the able-bodied range after extensive practice.
• Both amputees showed improved performance with probability-weighted regression for single-DOF tasks, mirroring able-bodied subject results.

Conclusions:

• Transradial amputees may require more training time to reach performance levels comparable to able-bodied individuals using intramuscular EMG-based prosthesis control.
• The comparative performance differences between linear and probability-weighted regression observed in able-bodied subjects are also applicable to amputee populations.
• Intramuscular EMG offers a viable pathway for advanced, multi-DOF prosthesis control in amputees.