An application of steady state visual evoked potential brain-computer interface as an augmentative alternative communication system for individuals with severe motor impairments

Abstract

PURPOSE: Tbis study will look at the feasibility of Steady State Visually Evoked Potential (SSVEP) brain-computer interfaces (BCI) as possible augmentative and alternative communication (AAC) systems for individuals who are severely disabled such as those with Locked-in Syndrome (LIS). The study intended to test whether there is a difference in BCI performance between healthy and impaired individuals and why. Specifically, the study focused on the operational competency, such as ocular motor function, ofthe impaired individuals as it relates to performance. Further, the study also attempted to explore the contributions of environmental distracts to performance. The results oftbis investigation will provide insights valuable for future BCI-AAC development and the potential for their acceptance by the AAC and LIS communities. METHODS: The study consisted of 12 healthy adults and 5 severely disabled adults presenting with 4 different neurological disorders. Tbis study consisted to two parts. The first part was an assessment ofthe communicative abilities ofthe impaired subjects. The assessment was conducted through a video recorded interview, from which communication rates were calculated and behavioral observations of each impaired subject's communicative behaviors were made with a focus on ocular motor behavior. The second part involved testing of the SSVEP BCI. All subjects performed selection tasks from a choice of four directions in the UDLR task. For each trial, the subject was prompted to attend to a specific SSVEP stimulus. Each stimulus was selected at random to flash at one of four frequencies (12, 13, 14, or 15Hz) (Lorenz, 2012). After 4 seconds, the BCI predicted the attended cue direction (Up, Down, Left, Right). If the prediction was correct, a "thumbs-up" feedback signal was shown to the subject; a "thumbs-down" was shown for incorrect predictions. The UDLR data collected for each trial consisted of a table with two columns: one column recorded the ground truth, which was the target direction, and one column recorded the decoded, or classified direction. Two additional columns were added. One column indicated whether the subject had any ocular motor impairment with a 1 or 0. A binary logistic regression was completed to investigate the main effect of age, subject group, and ocular motor impairment with respect to BCI accuracy. Additionally, observations regarding the affect of environmental distractions were also made. [TRUNCATED]