Evaluation of Linear Regression Simultaneous Myoelectric Control Using Intramuscular EMG

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Myoelectric prostheses traditionally require users to independently control individual muscles, which is difficult for amputees.
• Intramuscular electromyogram (EMG) signals offer rich data for prosthetic control.
• Decoding muscle coactivation patterns is key to advanced prosthetic function.

Purpose of the Study:

• To evaluate linear regression models for decoding intramuscular EMG signals.
• To assess the efficacy of these models in providing simultaneous myoelectric control of a virtual 3-DOF wrist/hand system.
• To compare this novel control method against conventional parallel dual-site control.

Main Methods:

• Linear regression control was tested in eight able-bodied subjects using a virtual Fitts' law task.
• Performance was compared to eight subjects using parallel dual-site control.
• Offline analysis examined the impact of training data on linear regression prediction accuracy.

Main Results:

• Both control systems showed similar overall performance.
• Linear regression excelled in tasks requiring all three degrees of freedom (DOFs).
• Parallel dual-site control was better for single-DOF tasks; linear regression users struggled with isolating movements but achieved easier simultaneous multi-DOF activation.

Conclusions:

• Linear regression myoelectric control using intramuscular EMG is a viable alternative for simultaneous 3-DOF wrist/hand prosthesis control.
• This method presents a trade-off between simultaneous control ease and individual DOF isolation.
• Different training data approaches may influence the controllability of linear regression systems.