Closed-Loop Control of Functional Electrical Stimulation Using a Selectively Recording and Bidirectional Nerve Cuff...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Sensory feedback is crucial for closed-loop control of functional electrical stimulation (FES) to restore movement in paralyzed limbs.
• Previous research utilized deep learning on spatiotemporal neural patterns from multi-contact nerve cuff electrodes for off-line classification.

Purpose of the Study:

• To demonstrate the feasibility of using deep learning for real-time neural signal classification in a closed-loop FES system.
• To enable sensory feedback for restoring movement in paralyzed limbs.

Main Methods:

• Acute in vivo experiments on 11 rats using a 64-channel nerve cuff electrode implanted on the sciatic nerve.
• Training a convolutional neural network (CNN) on spatiotemporal neural recordings for classifying hindpaw states (dorsiflexion, plantarflexion, heel prick).
• Implementing a rule-based closed-loop controller to generate ankle movements based on CNN output and neural stimulation.

Main Results:

• Successful closed-loop functional electrical stimulation demonstrated in 6 out of 11 subjects.
• Achieved 1-17 successful movement sequence trials per subject.
• Recorded 3-53 correct state transitions per trial, indicating effective neural signal decoding and FES control.

Conclusions:

• A convolutional neural network applied to multi-contact nerve cuff recordings is feasible for real-time closed-loop control of functional electrical stimulation.
• This approach shows promise for developing advanced neuroprosthetics and restoring motor function.