Creation of a biological sensorimotor interface for bionic reconstruction

Bionic reconstruction aims to restore hand function after amputation or severe injury where biological reconstruction is impossible. Advances in surgical techniques (targeted muscle reinnervation (TMR), osseointegration, regenerative peripheral nerve interfaces (RPNI)) and technology (implantable EMG electrodes, multi-electrode arrays, machine learning) have improved prosthetic control. However, the sophistication of robotic limbs outpaces the development of man-machine interfaces, leading to a mismatch between prosthetic capabilities and information transfer. Myoelectric prostheses, despite technological advancements, have only recently surpassed body-powered prostheses in popularity, partly due to the latter's straightforward control and immediate force/proprioceptive feedback. Restoring intuitive sensory feedback remains a significant challenge. Current methods, including sensory substitution, modality-matched feedback, and peripheral nerve stimulation, fail to replicate natural sensations and are often perceived as unpleasant. TMR and targeted sensory reinnervation (TSR) offer a new approach, where efferent fibers reinnervate a target muscle, and sensory fibers reinnervate overlying skin, creating biological communication pathways. While TSR shows promise, using glabrous skin grafts, with their higher mechanoreceptor density, could further enhance sensation restoration. This study investigated a biological sensorimotor interface using glabrous dermal skin graft transplantation in a targeted reinnervation rat model to create a bidirectional communication unit.

Numerous studies have explored restoring sensation in bionic limbs. Non-invasive techniques like sensory substitution and modality-matched feedback offer limited somatotopic accuracy and fail to replicate natural sensation. Invasive approaches, such as peripheral nerve stimulation, can provide afferent stimuli but often cause dysesthesia and paresthesia. While some studies show promise with proprioception elicitation via nerve stimulation, consistent, biomimetic activation for complex sensations remains challenging. TMR and TSR offer a more intuitive interface, with sensory fibers reinnervating the overlying skin, leading to modality-matched, somatotopically appropriate sensations. Combining TSR with non-invasive feedback systems may improve accuracy. While some studies utilize TSR with non-glabrous skin, this study focuses on glabrous skin grafts for potentially superior sensation restoration due to their higher innervation density and the avoidance of competition with native afferents. This approach also allows for strategic placement of grafts on superficial or deeper muscles based on the amputation stump. This strategy may provide the optimal biological sensory interface.

Thirty-two male Sprague-Dawley rats were divided into three groups: a skin graft group for tissue examination and electro-neurography (ENG) (n=13), a skin graft group for sequential retrograde labeling (n=9), and a control group without skin grafting (n=10). The ulnar nerve (UN), containing motor and sensory axons, was transferred to the long head of the biceps (LH). In the skin graft groups, a glabrous dermal skin graft from the hindlimb was transplanted onto the LH. A 12-week (13 weeks for retrograde labeling) follow-up period allowed for reinnervation. Immunofluorescence staining using antibodies against neurofilament (NF), S100, and myelin basic protein (MBP) was performed on thin and thick frozen sections to assess reinnervation. Sequential retrograde labeling with fast blue (FB) and Red Retrobeads™ (RB) quantified sensory reinnervation at the dorsal root ganglia (DRG) level. ENG measured afferent nerve activity in response to mechanical stimulation (Semmes-Weinstein monofilaments, vibration) of the skin graft (experimental groups) or muscle belly (control group). In the control group, tendon manipulation assessed proprioceptive responses. Statistical analyses (Kolmogorov-Smirnov test, unpaired t-test, Mann-Whitney U-test) compared axon counts, ENG amplitudes, and neuronal counts. Histology (H&E staining) assessed graft engraftment.

Surgical feasibility was confirmed, with all animals surviving the procedures without adverse events. Morphological analysis showed successful motor and sensory reinnervation, including reinnervation of Meissner corpuscles, Merkel disks, and the dermal plexus. Muscle spindles were also reinnervated, evidenced by large, myelinated nerve fibers. Sequential retrograde labeling quantified sensory reinnervation at the DRG level (approximately 13% of UN neurons contributed to dermal graft reinnervation) with minimal involvement of spinal cord neurons. ENG confirmed interface functionality, showing graded responses to monofilament stimulation and vibration in the skin graft group. Surprisingly, the control group also showed robust responses to touch and vibration, attributed to proprioceptive reinnervation of muscle spindles, confirmed morphologically via synaptophysin staining. Tendon manipulation further confirmed proprioceptive responses, indicating reinnervation of type Ia fibers. The unexpectedly robust responses in the control group highlight the potential importance of proprioceptive feedback in even the simplest nerve transfer setups. Comparison of superficial and perineural vibration responses ruled out movement artifacts as the source of tactile stimuli responses in the control group. Analyses of the fiber composition revealed a predominance of unmyelinated fibers in the dermal graft, possibly due to a mechanical barrier at the musculocutaneous junction. However, the presence of unmyelinated fibres in the glabrous skin graft does not seem to affect the quality of the tactile feedback generated by the interface.

This study presents a novel biological sensorimotor interface showing both tactile and proprioceptive feedback. Previous attempts at sensory restoration in prostheses using sensory substitution, modality-matched feedback, or peripheral nerve stimulation have limitations in terms of natural sensation and somatotopic precision. While these methods offer limited feedback, this work utilizes native cutaneous receptors and re-establishes both peripheral efferent and afferent pathways directly related to sensorimotor control. The proposed interface differs from existing approaches like C-RPNI and DS-RPNIs, offering a simpler surgical approach and the unique advantage of generating proprioceptive signals. The results show that both tactile and proprioceptive signals can be generated using the proposed interface. This interface is thus superior to other state-of-the-art biological interfaces as it combines both tactile and proprioceptive feedback, which may be crucial for improving the dexterity of bionic limbs. The combination of this interface with multiple nerve transfers and implantable multichannel EMG electrodes holds significant promise for enhancing man-machine interfaces in bionic reconstruction.

This study successfully established a biological sensorimotor interface in a rat model. Morphological and electrophysiological evidence confirms the reinnervation of skin, mechanoreceptors, and muscle spindles. This bidirectional interface demonstrates the potential for restoring both tactile and proprioceptive feedback in bionic limbs, paving the way for improved prosthetic control and user experience. Future research should focus on refining signal processing techniques to separate tactile and proprioceptive signals, investigating the long-term stability of the interface, and exploring clinical translation strategies.

This study used a rat model, which may not fully reflect human physiology. The relatively small sample size limits the statistical power of the findings. The study's focus on the rat model limits its immediate clinical translatability. The exact contribution of cutaneous and proprioceptive afferents to the recorded ENG signals remains to be fully elucidated. The observed variability in reinnervation across animals underscores the need for further research to optimize the surgical procedure and improve the consistency of reinnervation.