Proprioceptive Sensorimotor Integration with Neural Interfaces for Hand Prostheses

点击翻译按钮获取中文摘要

Current prosthetic options do not meet the needs of all Veterans with upper extremity limb loss. Chronically- implanted neural interfaces have shown promise for providing intuitive control and somatosensory feedback to prosthesis users. However, restoring proprioception, which is the sense of limb position and movement, via neural stimulation remains largely unexplored. Proprioception is critical for informing and modulating motor control in able-bodied individuals, and without it, prosthesis users must rely on visual feedback. The goal of this study is to understand the perception of proprioception from peripheral nerve stimulation (PNS) and its integration with prosthesis control. The central hypothesis is that sensorimotor integration will be best when both the proprioceptive inputs and the prosthesis control scheme match the underlying neural representations of sensorimotor processes in the extant body schema. Providing natural proprioception to Veterans with upper limb loss is the next critical step in the advancement of upper limb prosthetics toward having a hand again. Six participants with unilateral trans-radial or trans-humeral limb loss will be implanted with chronic neural and muscular interfaces. Composite Flat Interface Nerve Electrodes (C-FINEs) will be implanted around the residual median, ulnar, and radial nerves in the upper arm. Three trans-radial participants will be implanted with bipolar intramuscular electromyography (EMG) electrodes (IMs) in residual muscles in the forearm. Three trans-humeral amputees will receive Targeted Muscle Reinnervation (TMR) surgery at the time of implant and IMs will be placed in re-innervated muscles. Participants will attend laboratory testing sessions every 45 days for two years to complete experiments for the three specific aims of the study. In Aim 1, proprioceptive percepts elicited by peripheral nerve stimulation will be characterized. PNS-elicited perceived hand movements will be tracked through contralateral posture matching. Regression analyses will be performed to determine the contribution of various stimulation parameters on the resulting perceived hand kinematics. This information will be used to build a stimulation encoder to provide time-varying proprioceptive information. Paired agonist-antagonist PNS strategies will also be developed. The discriminability of agonist- only and paired agonist-antagonist strategies will be compared with psychometric tests. In Aim 2, the role of direct muscle activation from PNS on stimulation-elicited proprioceptive percepts will be examined. A nerve block using lidocaine injection will be performed on trans-radial participants to temporarily anesthetize the forearm muscles. Perceived movements and EMG activity will be compared before and after motor block for various PNS stimuli. For trans-humeral participants who will receive TMR, muscle re- innervation takes ~3 months. Proprioceptive percepts and EMG activity will be compared before and after re- innervation is complete. All participants will perform psychometric experiments assessing PNS stimuli, and discriminability will be compared for pulse frequencies above and below tetanic contraction frequency. In Aim 3, the integration of artificial proprioception with motor control will be assessed. In a virtual posture matching task, participants will control a virtual hand to achieve target postures using EMG signals recorded from the IMs. Two control paradigms will be compared: 1) position-based control, which aims to mimic natural hand posture control in able-bodied individuals, and 2) velocity-based control, which is typical for commercially-available prostheses. Performance will be compared with and without stimulation-evoked proprioceptive feedback. In a psychometric dissimilarity rating task, participants will rate the subjective dissimilarity of pairs of stimulation-evoked percepts. Some pairs will also contain voluntary hand motions along with the stimulation. Ratings will be compared between passive and active proprioceptive conditions to determine the effect of voluntary control on stimulation-evoked percepts. Surveys to assess embodiment and agency will also be compared before and after each research session.