FMRI-Guided Subdural Visual Motion BCI with Minimal Invasiveness

Abstract

Electrocorticography (ECoG) is a promising technology for high performance brain-computer interfaces (BCIs). To implement practical ECoG based BCIs, minimizing the invasiveness of the electrode implantation is critical. In this study, we advanced our recently proposed ‘N200 speller’ BCI paradigm that utilizes the attentional modulation of visual motion response. Non-invasive functional magnetic resonance imaging (fMRI) was employed to localize the visual motion processing regions. The subdural electrodes within these fMRI defined regions were associated with a negative deflection around 200 ms post-stimulus, and a power increase of the high gamma (60–140 Hz) frequency range around 100–500 ms post-stimulus, when the corresponding visual motion stimuli were attended. In subsequent BCI analyses, these electrodes showed top classification accuracies among all electrodes, suggesting the optimal locations for electrode implantation can be determined prior to surgery using fMRI imaging. Our findings demonstrate the feasibility of implementing a minimally invasive ECoG based N200 speller.