Efficient event related oscillatory pattern classification for EEG based BCI utilizing spatial brain dynamics

This paper features the spatial characteristics of the brain towards brain-computer interface (BCI) research. A study on motor imagery (MI) based BCI has been carried out and important implications are identified. Common Spatial Pattern (CSP) is applied to the EEG signals before proceeding to the classification. The primary focus of this research is to utilize the spatial dynamics of the brain to develop BCI with reduced number of electrodes which contribute to the motor imagery tasks with optimal impact. It is observed that computational cost can be reduced drastically by selecting channels from specific regions of interests (ROIs) of the brain without compromising the classification accuracy making BCI efficient. Here, we have reported the best classification accuracies 72.5% and 97.1% which are achieved for two subjects (`av' and `ay', respectively, in the dataset IVa in the BCI competition III) using less number of electrodes.