Year: 2017
Type: dissertation
Abstract: Brain computer interfaces (BCIs) can provide individuals with severe motor disorders a means of communication. This study aimed to determine whether visuospatial imagery could be used to signify intent in an electroencephalography (EEG)-based BCI. Eighteen healthy participants imagined stimuli in visual field quadrants while EEG was collected. A subset of participants used visuospatial imagery to control a character’s movement in four directions. Classifying rest against non-specific visuospatial imagery attained accuracies of 77±11.4%. Six participants exceeded chance when discriminating imagery in diagonally-opposing quadrants. These six could be predicted using a regression model that combined visuospatial perception and fatigue scores. The 4-class navigation accuracy was 48.3±18.8% (max 85%). Imagery in one hemifield corresponded to significant increases in alpha spindles in ipsilateral visual cortical regions. While further improvement is necessary to make the paradigm generalizable, visuospatial imagery shows promise for becoming a useful system for individuals in need of a BCI access technology.
Source: TSpace (University of Toronto)
Author: Filip Stojic
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
Funder: Bloorview Research Institute
OpenAlex ID: W3094162057