Robotic leg control with EMG decoding in an amputee with nerve transfers
During the described test patient’s knee-disarticulation amputation surgery, two nerve transfers were performed to prevent neuroma formation and allow re-innervation for use on a robotic leg.
The Electromyographic (EMG) signal in the residual limb was investigated using a cylindrical grid of 96 electrodes whilst the patient ‘performed’ movements of knee and ankle movement. Each movement showed a distinct signal pattern; therefore it was deemed a successful recognition system. The patient was then connected to a virtual limb by use of surface electrodes on the 8 natively innervated residual limb muscles and the 2 re-innervated muscle segments. This yielded a result of 92% accuracy with all configurable movements.
After 3 hours of accommodation time the patient was asked to perform different walking tasks with different pre-set impedances, the limb being controlled by both EMG and mechanical sensors compared to mechanical alone. The error rate dropped dramatically when the combined input was used, and no errors occurred that may cause the patient to fall.
There is feasibility in use of this prosthesis but the design needs further investigation and improvements to; ensure electrodes stay in full contact with skin, improve EMG pattern recognition and make it lighter, quieter, smaller and more reliable. > From: Hargrove, et al., N Engl J Med 369 (2013) 1237-1242. All rights reserved to Massachusetts Medical Society.
Visit the PubMed summary for more information or your article access.