Importance of the Quantitative Change of EEG Theta/Beta Ratio Between Preparation and Motor Imagery: Correlation with the Performance of Classification

Brain-computer interface (BCI) has been widely used to recognize the intention of users in the motor imagery-based (MI-based) paradigm. Although the previous decade has seen a significant increase in MI classification accuracy, MI-based BCI studies have mainly focused on the analysis of brain activity patterns either in the preparation phase or during MI tasks. This study proposes an approach for the evaluation of concentration for both during preparation and MI task, by monitoring attention level using before- and after-theta/beta ratio (TBR). In this endeavor, a total of 9 EEG recording was retrieved from BCI competition IV 2a dataset. Attention change index (ACI) was calculated before and after TBR. The classification performance was estimated from the EEGNet. For evaluating the correlation between ACI and accuracy, this study focused on the inter-hemispheric region (Fz, FCz, Cz) that is related to cognitive function. Data analyses showed that the ACI and classification accuracy had the highest positive correlation on FCz (r2=0.343). The findings of this study imply that changing of attention level during pre- and post-MI tasks could be an important factor in the performance of MI. Monitoring and analyzing the user attention levels might be highly beneficial for the MI classification.