Suitability of the dorsal column nuclei for a neural prosthesis: functional considerations

The brainstem dorsal column nuclei (DCN) may be an ideal target for a future neural prosthesis to restore somatosensation in tetraplegic patients. We aimed to investigate the functional and structural characteristics of the DCN, with the overarching goal of determining their suitability as a somatosensory neural prosthetic target. First, we review the neuroanatomy of the DCN and surrounding nuclei, including the cuneate, gracile, external cuneate, X, and Z nuclei, which together comprise the DCN-complex. We reveal that the DCN are not organised to only process and relay tactile information, as is commonly thought, but instead are a complex sensorimotor integration and distribution hub, with diverse projection targets throughout the hindbrain and midbrain. Next, we sought to show that somatosensory signals arriving in the DCN are reproducible, and that they carry decodable information about the location and quality of somatosensory stimuli, which we propose are necessary conditions for a potential somatosensory neural prosthetic target. We record somatosensory-evoked signals from various locations across the surface of the DCN in 8-week-old anaesthetised male Wistar rats. We characterised somatosensory-evoked DCN surface signals and demonstrated that they have robust and reproducible high-frequency and low-frequency features within and across animals. Using a machine-learning approach, we developed a metric for evaluating the relevance of machine-learning inputs to target outputs, which we coined feature-learnability. Using feature-learnability allowed us to determine the DCN signal features that were most relevant to peripheral somatosensory events, which facilitated very high accuracy prediction of the location and quality of somatosensory events, from small numbers of features. This thesis supports the DCN as a potential somatosensory neural prosthetic target by: i) showing DCN connectivity with sensorimotor targets essential for movement modulation in conscious and non-conscious neural pathways; ii) determining DCN signal features that are most relevant to peripheral tactile and proprioceptive events. New knowledge about the most relevant DCN signal features may inform the development of biomimetic stimulus patterns designed to artificially activate the DCN in future neural prosthetic devices for restoring somatosensory feedback