NOTICE: This event will be held via Zoom. The link will be placed below.

Zoom: https://umich.zoom.us/j/91278019863

Many diseases and injuries irreparably harm the brain or spinal cord and result in motor paralysis, widespread sensory deficits, and pain. Often, there are no treatments for these injuries, and therapies revolve around rehabilitation and adapting to the acquired deficits. In this work, we investigate brain machine interfaces (BMIs) as a future therapy to restore sensorimotor function, use BMIs to understand sensorimotor circuits, and use novel imaging algorithms to assess structural damage of somatosensory inputs into the brain.

Brain-controlled robotic arms have progressed rapidly from the first prototype devices in animals; however, these arms are often slow-moving compared to normal hand and arm function. In the first study, we attempt to restore higher-velocity movements during real-time control of virtual fingers using a novel feedforward neural network algorithm to decode the intended motor movement from the brain. In a non-human primate, the neural network decoder was compared with a linear decoder, the ReFIT Kalman filter (RFKF), that we believe represents the state-of-the-art in real-time finger decoding. The neural network decoder outperformed RFKF by acquiring more targets at faster velocities. This neural network architecture may also provide a blueprint for additional advances.

Somatosensory feedback from robotic arms is an important step to improve the realism and overall functioning. The use of somatosensory thalamus was investigated as a site of implantation for a sensory prosthesis in subjects undergoing awake deep brain stimulation surgery (DBS). In this study, electrical stimulation of the thalamus was performed using different stimulation patterns and the evoked sensations were compared. We found that the sensations evoked by bursting (a burst of pulses followed by a rest period) and tonic (regularly repeating pulses) stimulation were often in different anatomic regions and often with differing sensory qualities. These techniques for controlling percept location and quality may be useful in not only in BMI applications but also in DBS therapies to better relieve symptoms and avoid unwanted side effects.

Given the importance of sensory integration in motor functioning, the third study investigated the impact of a pharmacological perturbation on somatosensory content in primary motor cortex measured with Utah arrays implanted in two NHPs. Specifically, during continuous administration of nitrous oxide (N2O), somatosensory content was assessed by using the neural activity in primary motor cortex to classify finger brushings with a cotton-tip applicator. N2O degraded but did not eliminate somatosensory content in motor cortex. These findings provide insight into N2O mechanisms and may lead to further study of somatosensory afferents to motor cortex.

A debilitating facial pain syndrome, called trigeminal neuralgia (TN), is thought to be caused by vascular compression of the sensory root that provides somatosensory feedback from the face. In this final study, magnetic resonance diffusion tensor imaging was used to assess the structural damage of this sensory root. In a retrospective manner, we developed and tested an algorithm that predicted the likelihood of pain relief after surgical treatment of TN. This algorithm could help select patients for surgery with the best chance for pain relief.

Together, these studies advance BMI technologies that attempt to restore realistic function to those with irreparable damage to sensorimotor pathways. Furthermore, using BMIs and novel imaging, this work provides a better understanding of sensorimotor circuits and how sensory pathways can be damaged in disease states.

Co-Chairs: Parag G. Patil and Cynthia A. Chestek