Reliable Prosthetic Hand and Wrist Control Using Regenerative Peripheral Nerve Interfaces (RPNIs) and Implanted Electrodes
Alex K. Vaskov, PhD1; Christina Lee, MS1; Philip P Vu, PhD1; Dylan M Wallace, BS2; Alicia J Davis, MPA, CPO1; Theodore A Kung, MD3; Cynthia A Chestek, PhD4; Paul S Cederna, MD5; Stephen WP Kemp, PhD1
1University of Michigan, Ann Arbor, MI; 2Univeristy of Michigan, Ann Arbor, MI; 3Section of Plastic & Reconstructive Surgery, University of Michigan, Ann Arbor, MI; 4Biomedical Engineering, University of Michigan, Ann Arbor, MI; 5Plastic Surgery, University of Michigan, Ann Arbor, MI
Introduction: State-of-the-art prosthetic limbs can actuate multiple hand and wrist movements, but the lack of reliability and intuitiveness leads to user dissatisfaction and device abandonment. Intramuscular electrodes can improve control fidelity. However, amputation limits the number of muscles available to provide signals for complex prosthetic functions. In clinical trials, the Regenerative Peripheral Nerve Interface (RPNI) has been shown to amplify efferent nerve action potentials in lieu of missing muscles. The purpose of this study was to determine if these amplified RPNI signals can be used to achieve reliable prosthetic hand and wrist control in individuals with upper limb amputation.
Materials and Methods: RPNIs were created by implanting a severed peripheral nerve into a free muscle graft. Three participants with transradial amputations (P1, P2, and P3) had intramuscular bipolar electrodes surgically implanted into RPNIs on their median and ulnar nerves, and 5-8 residual muscles. The percutaneous electrodes were connected to a real-time computer.
Results: To date, signal-to-noise ratios (SNRs) have been measured approximately monthly across 267 days for P1 (12 sessions), 1240 days for P2 (37 sessions), and 335 days for P3 (12 sessions). Thumb flexion evoked the largest amplitude response from the median nerve RPNIs with average (mean±std) SNRs of 160±69.0 (P1), 25.3±8.8 (P2), and 280±35.3 (P3). Small finger flexion or wrist flexion evoked the largest response from the ulnar RPNIs with average SNRs of 35.8±11.0 (P1), 32.7±11.9 (P2), 14.7±2.5 (P3). There have been no significant declines in the SNR of RPNI signals over time (p > 0.05, F-test).
Patients used a real-time pattern recognition system to control commercial multi-grip prosthetic hands with electromyography from RPNIs and residual muscles. Consistent with previous work, control parameters did not require re-calibration between experiment sessions. P1 used fist, pinch, and point to complete a multi-grip reach and place task with 89.9% accuracy. P3 used hand open and close along with wrist pronation and supination to complete pouring, jar opening, and object moving of the Southampton Hand Assessment Procedure with a linear index of functionality of 69.7 (best trial, range 0-100). P2 used fist, pinch, point and active wrist rotation to complete a multi-grip coffee making task with a grip selection accuracy of 90%.
Conclusion: RPNIs produced large amplitude motor signals with no signal degradation over time, enabling all three participants to control multiple prosthetic functions accurately and reliably. Clinical translation of this technology will increase prosthetic functionality and reduce device abandonment.
Back to 2023 Abstracts