Improving the Cross-Subject Performance of the ERP-Based Brain–Computer Interface Using Rapid Serial Visual Presentation and Correlation Analysis Rank

• Published in: Frontiers in human neuroscience Vol. 14; p. 296
• Main Authors: Liu, Shuang, Wang, Wei, Sheng, Yue, Zhang, Ludan, Xu, Minpeng, Ming, Dong
• Format: Journal Article
• Language: English
• Published: Lausanne Frontiers Research Foundation 31.07.2020; Frontiers Media S.A
• Subjects: Accuracy; Auditory evoked potentials; Brain; brain–computer interface; Classification; Computer applications; Correlation analysis; cross-subject; Electrodes; Electroencephalography; event-related potential; Event-related potentials; Experiments; Genetic algorithms; Implants; Laboratories; Neuroscience; Principal components analysis; rapid serial visual presentation; Signal processing
• ISSN: 1662-5161; 1662-5161
• DOI: 10.3389/fnhum.2020.00296

The brain-computer interface (BCI) is a system that is designed to provide communication channels to anyone through a computer. Initially, it helped the disabled, but actually had been proposed a wider range of applications. However, the cross-subject recognition in BCI systems is difficult to break apart from the individual specific characteristics, the unsteady characteristics and the environmental specific characteristics, which also makes it difficult to develop high reliable and high stabile BCI systems. Rapid Serial Visual Presentation (RSVP) is one of the most recent spellers with a clean, unified background and a single stimulus, which may evoke Event-Related Potential (ERP) patterns with less individual difference. In order to build a BCI system that allows new users to use it directly without calibration or with less calibration time, RSVP was employed as evoked paradigm, then Correlation Analysis Rank (CAR) algorithm was proposed to improve the cross-individual classification and simultaneously use as less training data as possible. Fifty-eight subjects took part in the experiments. The flash stimulation time is 200ms and the off time is 100ms. The P300 component was locked to the target representation by time. The results showed that RSVP could evoke more similar ERP patterns among subjects compared with matrix paradigm. Then the included angle cosine was calculated and counted for averaged ERP waveform between each two subjects and the threshold value was set to 0.5. The average matching number of all subjects was 6 for matrix paradigm, while for RSVP paradigm, the matching number range was 20, which was more than three times as much larger, indicating that ERP waveforms evoked by RSVP paradigm produced smaller individual differences, which is more favorable for cross-subject classification and recognition. Receiver operating characteristic (ROC) curve value were computed and compared using CAR and traditional random selection. The results showed that the proposed CAR got significantly better performance than traditional random selection, and got the best AUC value of 0.8, while traditional random selection only achieved 0.65. These encouraging results suggest that with proper evoked paradigm and classification methods, it is feasible to get stable performance across subjects for ERP-based BCI.