A hybrid EOG-P300 BCI with dual monitors

Abstract

P300 brain computer interface (BCI) is one of the most widely studied BCI paradigm. It detects the specific wave form, produced in response to task-relevant stimuli. In visual stimulus based P300 BCI system, the subjects select the target item by gazing at it. With regarding that the accompanying eye movements to change a gazing object can evoke the electrooculography (EOG) responses, P300-based BCI and EOG-based gaze tracking system can be beneficially integrated to improve the performance. Based on this idea, we investigate novel hybrid EOG-P300 BCI system with dual monitors. From the ordinary P300 interface, we split the menu items into dual monitors. The system analyzes EOG signals to find which monitor was focused by the subjects, then in the monitor, the P300 system identifies the item focused by the subjects. With reducing the number of items in a screen, we can reduce by almost half the time required to select a single item. To evaluate our hybrid P300 system, we computed classification accuracy and practical bit rate (PBR), and the system was compared with the conventional P300 system. The hybrid system scored classification accuracy of 80% and 0.5556 PBR. When we compared the hybrid BCI with the ordinary P300 BCI, the classification accuracies of the systems were almost same that is p-value between two groups of accuracies was estimated to be 0.3217 by one tailed t-test. In PBR, the hybrid BCI showed 70% higher PBR than the conventional P300 BCI. These evaluation results proved possibility of the hybrid BCI for practical use with high speed and reliability.