METHODS: The RPNI consists of a free muscle graft implanted on the end of a transected nerve fascicle. Two macaques RPNIs implanted (n=3/macaque) in the forearm. Intramuscular electromyography (EMG) electrodes were implanted in each macaque RPNI. Macaques were trained to perform index finger movements to acquire virtual targets on a computer screen.5 Finger position was recorded via a flex sensor on the index finger. The human surgery was completed to treat neuroma pain. Peripheral nerves of a low forelimb amputee were divided proximal to existing neuromas on three nerves, and small pieces of skeletal muscle were freely transferred to the proximal residual end of each peripheral nerve. The human had fine wire electrodes temporarily placed in the RPNIs under ultrasound guidance. Voluntary RPNI EMG that represented control was sensed by the implanted electrodes.
RESULTS: With continuous EMG decode using 10-fold cross-validation, the resulting predicted finger position had a, correlation coefficient ρ=0.82 between predicted and true finger positions for Macaques. The EMG decode correctly classified 97.7% of movements (out of 261 total movements). The human ulnar nerve RPNIs were able to control thumb key pinch to first finger and thumb to little finger. Thumb abduction trials were 96% correct. At harvest macaque RPNIs were well vascularized; RPNI muscle fibers varied in diameter with some fibers showing continuing regeneration after implantation for 1 year.
CONCLUSIONS: Macaques voluntarily controlled virtual finger movements with nerve signals transferred through implanted RPNIs. The human controlled an advanced myoelectric prosthetic thumb with RPNIs implanted in the ulnar nerve RPNI. Voluntary RPNI control of a myoelectric hand was validated in both macaque and human models.