The effect of selecting different time interval as event and type of phenomenon in performance of SBCI systems

Self-paced Brain Computer Interface (SBCI) is an important class of BCI systems which has not received enough attention yet. As all brain researchers know certain events, such as motor imagery, can elicit event-related synchronization/ de synchronization (ERD/ERS) of neurons in the brain which manifest themselves in terms of band power changes in brain signals. The activation/deactivation of brain´s cortical neural systems during a motor activity changes the complexity or randomness of spontaneous EEG and can be quantified accurately with fractal dimension & entropy. In this paper, we introduce two efficient factors in improving the performance of SBCI systems that are :1) determining a proper time interval in EEG signal as an event, we select this time interval which the phenomena are more able to indicate the brain activity, 2) selecting appropriate phenomenon in two ways. In first way the power of each phenomenon (ERD or ERS) is variable in different subjects and it is important to choose one which is appropriate in application, in second way we select both of them with different method. With suitable selection of this variables, performance of SBCI and TPR increase and FPR decreases. This factors are also different for all the subjects and related to type of method. Thus it is subject-independent and must be select manually.