Towards Control of a Transhumeral Prosthesis with EEG Signals

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Robotics

Background:

• Controlling multi-degree-of-freedom (DoF) transhumeral prostheses is challenging due to limited residual muscle signal diversity.
• Existing control methods for advanced prosthetics often fall short in providing intuitive and precise functionality for amputees.

Purpose of the Study:

• To propose and evaluate an electroencephalogram (EEG)-based hierarchical two-stage control strategy for multi-DoF transhumeral prostheses.
• To address the limitations of current prosthetic control systems by leveraging brain-computer interfaces.

Main Methods:

• Utilized a hierarchical two-stage approach employing EEG signals to decode motion intentions.
• Implemented neural network and k-nearest neighbor classifiers to identify arm reaching or hand lifting intentions.
• Employed neural-network-based classifiers trained on healthy subject data to estimate elbow and hand endpoint motion, comparing predictions with residual limb motion.

Main Results:

• Demonstrated the feasibility of the proposed EEG-based method for achieving multi-DoF control of transhumeral prostheses.
• Experimental results confirmed the effectiveness of the hierarchical two-stage approach in a proof-of-concept study with healthy subjects.

Conclusions:

• The developed EEG-based hierarchical control strategy shows promise for improving the functionality and intuitiveness of transhumeral prostheses.
• This approach offers a potential pathway to greater freedom and mobility for amputees using advanced robotic limbs.