再生末梢神経インターフェース（RPNI）と埋め込み電極による手と手首の義肢のオンライン制御の向上

Upper limb amputation severely limits daily activities and independence. Current prosthetic control methods often rely on surface electromyography (sEMG), which suffers from low signal quality and limited functionality. This study investigates whether implanted electrodes in neuromuscular Regenerative Peripheral Nerve Interfaces (RPNIs) and residual innervated muscles can provide stable and high-quality control signals to improve dexterous prosthetic hand and wrist function. Approach: Neuromuscular RPNIs (i.e. made from a free muscle graft wrapped around a nerve branch, no sensory tissue) were created and intramuscular EMG (iEMG) electrodes were implanted into the RPNIs and residual innervated muscles for two individuals with upper-limb amputation. EMG signals were recorded from both surface (sEMG) and implanted (iEMG) electrodes and used to control a virtual prosthetic hand + wrist in real time. Performance was assessed through multiple degrees-of-freedom (DoF) control tasks, comparing iEMG against conventional sEMG. Main Results: Implanted electrodes demonstrated high signal-to-noise ratios and long-term stability, enabling independent and simultaneous control of multiple hand + wrist DoFs. Participants achieved faster, more accurate, and more reliable control using iEMG-based control compared with sEMG-based systems, based on classification accuracy and trial success rate. Importantly, we find that the ability to control wrist rotation reduces total body compensations when performing a functional assessment (Coffee Task), and implanted electrodes greatly reduced task completion times compared to surface electrodes when wrist rotation was added as an additional control movement. Significance: In this study, we demonstrate that RPNIs and iEMG electrodes enable significantly more accurate and stable prosthetic control of hand and wrist movements compared to surface electrodes, especially during dynamic arm movements. These findings suggest that implanted electrodes offer meaningful advantages over sEMG for achieving more intuitive and reliable control of upper limb prostheses in real-world conditions.