Investigations into Brain-Computer Interfacing for Stroke Rehabilitation

A stroke is the loss of brain function following the cessation of blood supply to a region of the brain caused by either a blockage or haemorrhage in the vasculature. It is a leading cause of death worldwide but survival rates have increased significantly in the past 25 years with recent estimates putting the number of worldwide stroke survivors at 33 million. Stroke survivors live with lasting effects such as limb weakness, limb paralysis, loss of speech, loss of comprehension and other neurological disorders. The purpose of stroke rehabilitation is to return the sufferer to as normal a life as possible. Traditional methods for this involve mass practice of the affected function to provoke improvement, acquisition of compensatory skills and adaptation to residual post-stroke disability. Recently, however, brain computer interfaces (BCI) have emerged as a technology which may have impact in augmenting traditional approaches, particularly for motor deficits. In this context, BCI provides a means for closing the sensorimotor loop and driving neuroplastic processes to enhance recovery. A BCI is a system for translating measured brain activity into control signals for an external device, such as a computer or machine. Rehabilitation BCI attempts to use such a device to encourage positive neurorehabilitation in the stroke survivor, to return or strengthen lost or diminished function. This thesis describes concerted work to improve the current state and future prospects of rehabilitation BCI. In particular, BCIs which use electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) to measure brain activity are the focus of these efforts. EEG and fNIRS are relatively inexpensive, easy-to-use and portable brain measurement/ imaging systems compared to other brain imaging methods commonly found in hospital settings, such as functional magnetic resonance imaging (fMRI), positron emission tomography (PET) or magnetoencephalography (MEG). These advantages motivate this research in the hope that at-home stroke rehabilitation becomes widespread and the accepted method of stroke rehabilitation. Investigations described here include the design and development of a novel fNIRS imaging method, a novel fNIRS synthetic data generation algorithm, a novel hybrid fNIRS/EEG measurement system, a novel portable EEG biofeedback BCI, a substantial investigation into the effect of stroke on EEG BCI operation and performance, and an investigation into potential biomarkers for neurorehabilitation based on BCI parameters and scalp EEG. These investigations, based on measurements of both healthy and stroke affected brain activity, have lead to the advancement of EEG and fNIRSbased rehabilitation BCI technology