NEC seminar
Presenter: Brandon Ruszala, University of Rochester
Host: Prof. Bolu Ajiboye
Title: Interfacing with the Cortical Reach-to-Grasp Network using Low-Amplitude Intracortical Microstimulation
Abstract: Movement is the primary way people interact with the world. Injuries to the nervous system that disrupt a person’s ability to move (e.g., losing a limb or paralysis) can be devastating. However, the cortical regions that control movement often remain intact and functional offering an interesting potential treatment – communicating with those still-functional brain regions to control a machine and bypass the injuries. In other words, establishing a brain-machine interface (BMI). Focusing on upper-extremity BMIs, robotic arms can be successfully controlled by decoding neurons from motor cortex and improved by delivering sensory feedback to somatosensory cortex with intracortical microstimulation (ICMS). However, controlling BMIs with the speed, accuracy, and precision of natural movements made with native limbs remains a challenge.
This dissertation identifies several features of the cortical motor system that could be leveraged to refine BMI control. First, we show that neurons in motor cortex encode instruction modality – a non-kinematic feature that may constitute noise for BMIs trained to decode movement kinematics from those neurons. Accounting for such non-kinematic variation in future decoding algorithms may allow for more accurate decoding of movement parameters from that neural activity. Second, we show low-amplitude ICMS can modulate neurons in distant cortical regions. Initially, we show ICMS delivered in primary somatosensory cortex modulates the activity of neurons across wide territories in primary motor cortex and ventral premotor cortex. Subsequently, we show distant modulation effects can be produced over even greater spatial scale across the entire cortical reach-to-grasp network. For BMIs that concurrently stimulate neurons in some cortical regions while decoding neurons in others, modulation produced by the stimulation in decoded neurons can hinder decoder performance. Incorporating information about distant modulation effects into decoding algorithms could mitigate that hindrance. Finally, we explored the cortical reach-to-grasp network for novel regions in which information can be delivered using ICMS. We found that the ventral premotor cortex and the anterior intraparietal area were effective regions for delivering information, whereas the dorsal premotor cortex and dorsal posterior parietal cortex were ineffective. Future BMIs might deliver more complex information to the brain via those former regions, expanding the bidirectional brain-machine interface.