Publisher
The University of Arizona.Rights
Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author.Abstract
Spinal cord injury leaves high level tetraplegics unable to control their upper extremities. Functional electrical stimulation can be used to artificially activate muscles. Restoration of movement using such functional electrical stimulation thus far has been limited to a small set of simple movements due to the difficulty in identifying stimulation patterns for more complex behaviors. Inspired by the natural control of muscles, the work in this dissertation uses muscle activity and movements recorded in healthy subjects to estimate the relationship between them using artificial neural networks. To gather the muscle activity needed to train the artificial neural network, a large-scale intramuscular electrode system was developed. Muscle activity and movement kinematics recorded during reaching behavior in non-human primates were used to train the artificial neural networks. The trained networks were used to identify the stimulation patterns needed to evoke a wide range of movements. Intramuscular electrodes used to record muscle activity were then used to stimulate the muscles using the identified stimulation patterns. The movements generated from these stimulation patterns matched the intended movement significantly better than a stimulation pattern shuffled across muscles. Overall, the intramuscular electrode system developed here provided stable recording and stimulation over months. We also developed and compared non-invasive systems needed to extract movement intention that would serve as an input to drive our ANN-based functional electrical stimulation. These systems relied on head movement or voice, both of which are available to high-level tetraplegics. As a stand-in for an arm controlled with functional electrical stimulation, a robotic arm was used. Human subjects used head movements to control robot arm position, robot arm velocity, or voice to perform a 3D pointing task. Head control with position mapping worked significantly better than the other methods. The work presented here demonstrates significant advances toward controlling functional electrical stimulation for unrestricted upper limb movements.Type
textElectronic Dissertation
Degree Name
Ph.D.Degree Level
doctoralDegree Program
Graduate CollegeNeuroscience