Decoding Native Cortical Representations for Flexion and Extension at Upper Limb Joints Using Electrocorticography.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Traditional brain-machine interfaces (BMI) focus on endpoint control for prosthetics.
• A joint-centered approach may offer more intuitive control over limb movements.
• Electrocorticography (ECoG) offers high-resolution neural recordings.

Purpose of the Study:

• To investigate the feasibility of discriminating upper limb joint flexion/extension using ECoG.
• To evaluate a joint-centered control strategy for BMI.
• To assess decoding accuracy across different ECoG array densities.

Main Methods:

• Utilized ECoG recordings from sensorimotor cortex in four subjects.
• Subjects performed cued finger, wrist, and elbow flexion/extension movements.
• Trained linear models on high-gamma power (70-110 Hz) for movement classification.

Main Results:

• Offline decoding models achieved accuracies ranging from 62% to 83%.
• High-gamma power modulations successfully discriminated between flexion and extension.
• Widespread ECoG coverage enabled sampling of cortical representations for multiple joints.

Conclusions:

• A joint-centered approach is feasible for BMI control of upper limb movements.
• ECoG recordings from sensorimotor cortex can decode individual joint actions.
• This approach holds promise for developing more comprehensive and intuitive prosthetic limb control.