Fractal Dimension for Detection of ERD/ERS Patterns in Asynchronous Brain Computer Interface

Detection of motor-related events is the key issue in asynchronous brain-computer interface design. In this study we exploited for the first time Katz's fractal dimension for detection of motor related changes characterized by ERD/ERS patterns in electroencephalogram signal. Our observation was that the activation/deactivation of brain's cortical neural systems, during occurrence of motor activity, changes the complexity or randomness of spontaneous EEG and can be quantified accurately with fractal dimension. Furthermore, we applied a cross-correlation template matching (CCTM) method on the extracted features to combine the energy changes of both ERD and ERS patterns. This combination boosts the system capability and rapidity in motor activity detection. Evaluations of our proposed method shows advantage compared to entropy features extracted in [2], and reveals true positive rates of 90%-100% with corresponding false positive rates of 16.12%-0%, respectively.