Date of Award
Doctor of Philosophy (PhD)
Improving the recovery of lost motor function in hemiplegic chronic stroke survivors is a critical need to improve the lives of these patients. Over the last several decades, neuroprosthetic systems have emerged as novel tools with the potential to restore function in a variety of patient populations. While traditional neuroprosthetics have focused on using neural activity contralateral to a moving limb for device control, an alternative control signal may be necessary to develop brain-computer interface (BCI) systems in stroke survivors that suffer damage to the cortical hemisphere contralateral to the affected limb. While movement-related neural activity also occurs in the hemisphere ipsilateral to a moving limb, it is uncertain if these signals can be used within BCI systems. This dissertation examines the motor activity ipsilateral to a moving limb and the potential use of these signals for neuroprosthetic applications in chronic stroke survivors. Patients performed three-dimensional (3D) reaching movements with the arm ipsilateral to an electrocorticography (ECoG) array in order to assess the extent of kinematic information that can be decoded from the cortex ipsilateral to a moving limb. Additionally, patients performed the same task with the arm contralateral to the same ECoG arrays, allowing us to compare the neural representations of contralateral and ipsilateral limb movements. While spectral power changes related to ipsilateral arm movements begin later and are lower in amplitude than power changes related to contralateral arm movements, 3D kinematics from both contralateral and ipsilateral arm trajectories can be decoded with similar accuracies. The ability to decode movement kinematics from the ipsilateral cortical hemisphere demonstrates the potential to use these signals within BCI applications for controlling multiple degrees of freedom. Next we examined the relationship between electrode invasiveness and signal quality. The ability to decode movement kinematics from neural activity was significantly decreased in simulated electroencephalography (EEG) signals relative to ECoG signals, indicating that invasive signals would be necessary to implement BCI systems with multiple degrees of freedom. For ECoG signals, the human dura also causes a significant decrease in signal quality when electrodes with small spatial sizes are used. This tradeoff between signal quality and electrode invasiveness should therefore be taken into account when designing ECoG BCI systems. Finally, chronic stroke survivors used activity associated with affected hand motor intentions, recorded from their unaffected hemisphere using EEG, to control simple BCI systems. This demonstrates that motor signals from the ipsilateral hemisphere are viable for BCI applications, not only in motor-intact patients, but also in chronic stroke survivors. Taken together, these experiments provide initial demonstrations that it is possible to develop BCI systems using the unaffected hemisphere in stroke survivors with multiple degrees of freedom. Further development of these BCI systems may eventually lead to improving function for a significant population of patients.
Eric C Leuthardt
Dennis L Barbour, Maurizio Corbetta, Catherine E Lang, Daniel W Moran, Kurt A Thoroughman