AUTHOR=Geronimo Andrew , Kamrunnahar Mst , Schiff Steven J. TITLE=Single Trial Predictors for Gating Motor-Imagery Brain-Computer Interfaces Based on Sensorimotor Rhythm and Visual Evoked Potentials JOURNAL=Frontiers in Neuroscience VOLUME=10 YEAR=2016 URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2016.00164 DOI=10.3389/fnins.2016.00164 ISSN=1662-453X ABSTRACT=

For brain-computer interfaces (BCIs) that utilize visual cues to direct the user, the neural signals extracted by the computer are representative of ongoing processes, visual evoked responses, and voluntary modulation. We proposed to use three brain signatures for predicting success on a single trial of a BCI task. The first two features, the amplitude and phase of the pre-trial mu amplitude, were chosen as a correlate for cortical excitability. The remaining feature, related to the visually evoked response to the cue, served as a possible measure of fixation and attention to the task. Of these three features, mu rhythm amplitude over the central electrodes at the time of cue presentation and to a lesser extent the single trial visual evoked response were correlated with the success on the subsequent imagery task. Despite the potential for gating trials using these features, an offline gating simulation was limited in its ability to produce an increase in device throughput. This discrepancy highlights a distinction between the identification of predictive features, and the use of this knowledge in an online BCI. Using such a system, we cannot assume that the user will respond similarly when faced with a scenario where feedback is altered by trials that are gated on a regular basis. The results of this study suggest the possibility of using individualized, pre-task neural signatures for personalized, and asynchronous (self-paced) BCI applications, although these effects need to be quantified in a real-time adaptive scenario in a future study.