PhD Student, University of Alberta
I develop adaptable control strategies to restore walking after neural injury using a spinal implant
The overall goal of my work is to develop an intervention to improve mobility after an incomplete spinal cord injury (SCI). The most common type of SCI is incomplete, where some weak ability to move remains, and the functional deficits vary from person-to-person. Restoring walking is a top priority for people with paraplegia and is a central research and clinical goal. Intraspinal microstimulation (ISMS) is a novel approach developed in our lab for restoring walking after complete paralysis due to SCI. This method involves the implantation of fine, hair-like microwires into a small region of the spinal cord that houses connections to leg muscles. Low levels of electrical current are passed through these wires to activate these areas to produce functional leg movements.The procedure is safe, feasible and stable for long periods of time. As part of my work, I will develop strategies to control walking after an incomplete SCI. Specifically, I will develop control strategies for ISMS to dynamically adapt to each individual’s injury and augment remaining function using machine learning. By using ISMS to restore walking daily, it is probable that after removing the stimulation, voluntary control, coordination, and balance will improve. Over time, the individual may be less reliant on the assistance provided by ISMS to walk. My work will not only restore lost function due to injury, but also assist the spinal cord in forming new effective connections. Adaptive control strategies will allow for patient-specific rehabilitation, which is necessary as no two SCIs are alike. My work can also be expanded to restore walking in other neural injuries or diseases, including stroke and traumatic brain injury.
Abstract: Neuroprosthetic approaches have tremendous potential for the treatment of injuries to the brain and spinal cord by inducing appropriate neural activity in otherwise disordered circuits. Substantial work has demonstrated that stimulation applied to both the central and peripheral nervous system leads to immediate and in some cases sustained benefits after injury. Here we focus on cervical intraspinal microstimulation (ISMS) as a promising method of activating the spinal cord distal to an injury site, either to directly produce movements or more intriguingly to improve subsequent volitional control of the paretic extremities. Incomplete injuries to the spinal cord are the most commonly observed in human patients, and these injuries spare neural tissue bypassing the lesion that could be influenced by neural devices to promote recovery of function. In fact, recent results have demonstrated that therapeutic ISMS leads to modest but sustained improvements in forelimb function after an incomplete spinal cord injury (SCI). This therapeutic spinal stimulation may promote long-term recovery of function by providing the necessary electrical activity needed for neuron survival, axon growth, and synaptic stability.
Pub.: 01 Mar '14, Pinned: 29 Jun '17
Abstract: The term "nexting" has been used by psychologists to refer to the propensity of people and many other animals to continually predict what will happen next in an immediate, local, and personal sense. The ability to "next" constitutes a basic kind of awareness and knowledge of one's environment. In this paper we present results with a robot that learns to next in real time, predicting thousands of features of the world's state, including all sensory inputs, at timescales from 0.1 to 8 seconds. This was achieved by treating each state feature as a reward-like target and applying temporal-difference methods to learn a corresponding value function with a discount rate corresponding to the timescale. We show that two thousand predictions, each dependent on six thousand state features, can be learned and updated online at better than 10Hz on a laptop computer, using the standard TD(lambda) algorithm with linear function approximation. We show that this approach is efficient enough to be practical, with most of the learning complete within 30 minutes. We also show that a single tile-coded feature representation suffices to accurately predict many different signals at a significant range of timescales. Finally, we show that the accuracy of our learned predictions compares favorably with the optimal off-line solution.
Pub.: 08 Jun '12, Pinned: 29 Jun '17
Abstract: The goal of this study was to decode sensory information from the dorsal root ganglia (DRG) in real time, and to use this information to adapt the control of unilateral stepping with a state-based control algorithm consisting of both feed-forward and feedback components.In five anesthetized cats, hind limb stepping on a walkway or treadmill was produced by patterned electrical stimulation of the spinal cord through implanted microwire arrays, while neuronal activity was recorded from the DRG. Different parameters, including distance and tilt of the vector between hip and limb endpoint, integrated gyroscope and ground reaction force were modelled from recorded neural firing rates. These models were then used for closed-loop feedback.Overall, firing-rate-based predictions of kinematic sensors (limb endpoint, integrated gyroscope) were the most accurate with variance accounted for >60% on average. Force prediction had the lowest prediction accuracy (48 ± 13%) but produced the greatest percentage of successful rule activations (96.3%) for stepping under closed-loop feedback control. The prediction of all sensor modalities degraded over time, with the exception of tilt.Sensory feedback from moving limbs would be a desirable component of any neuroprosthetic device designed to restore walking in people after a spinal cord injury. This study provides a proof-of-principle that real-time feedback from the DRG is possible and could form part of a fully implantable neuroprosthetic device with further development.
Pub.: 10 Aug '13, Pinned: 29 Jun '17