부산대학교 도서관

Setting low-cost brain computer interfaces (BCIs) has been a topic of interest in developing countries. There is a variety of EEG equipment, and mathematical techniques that can help achieve this goal, but some of these techniques may have some flaws by their own. A low complexity alternative using an OpenBCI optimized equipment and based on two mathematical techniques here is discussed for an ongoing develoment of a low-cost BCI project. The selected techniques for data analysis inspection are linear discriminant analysis (LDA) and multivariate synchronization index (MSI). The procedure will be shown from the basics, the OpenBCI optimized equipment is validated offline against a g.Nautilus EEG and the performance of the techniques is shown individually. Therefore, it is discussed the possibility of combining both feature extraction techniques (LDA and MSI) for steady state visual evoked potentials (SSVEP), in order to achieve a concept with better performance for an SSVEP BCI.