AUTHOR=Won Kyungho , Kwon Moonyoung , Jang Sehyeon , Ahn Minkyu , Jun Sung Chan 

TITLE=P300 Speller Performance Predictor Based on RSVP Multi-feature

JOURNAL=Frontiers in Human Neuroscience

VOLUME=Volume 13 - 2019

YEAR=2019

URL=https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2019.00261

DOI=10.3389/fnhum.2019.00261

ISSN=1662-5161

ABSTRACT=Brain-computer interfaces (BCIs) have been developed so that people can control computers or machines through their brain activity without limb movements. The P300 speller is one of the BCI applications most commonly used as it is very simple and reliable and may achieve relatively higher accuracy compared to other types of BCIs. However, like other BCIs, the P300 speller still has performance variation issues in practical use, for example, selecting the best tradeoff between spelling accuracy and information transfer rate (speed). Therefore, seeking correlates of the P300 speller’s performance and predicting their performance is necessary to improve the P300 speller. In this work, we investigated the correlations between rapid serial visual presentation (RSVP) task features and the P300 speller performance. Fifty-five subjects participated in the RSVP and conventional matrix P300 speller tasks, and RSVP behavioral features and electroencephalography features were compared to the P300 speller performance. We found that several ERP features and behavioral features of the RSVP were correlated with the P300 speller’s offline binary classification accuracy. Using these RSVP features, we proposed a simple multi-feature performance predictor (r = 0.53, p < 0.0001) that outperformed any single feature performance predictor, including the conventional RSVP T1% predictor (r  = 0.32, p < 0.05). This result demonstrates that selective multi-features can predict BCI performance better than only a single feature.