EEG-Based Fractal Analysis of Different Motor Imagery Tasks using Critical Exponent Method

— The objective of this paper is to characterize the spontaneous electroencephalogram (EEG) signals of four different motor imagery tasks and to show hereby a possible solution for the present binary communication between the brain and a machine or a brain-computer interface (BCI). The processing technique used in this paper was the fractal analysis evaluated by the critical exponent method (CEM). The EEG signal was registered in 5 healthy subjects, sampling 15 measuring channels at 1024 Hz. Each channel was preprocessed by the Laplacian space ltering so as to reduce the space blur and therefore increase the space resolution. The EEG of each channel was segmented and its fractal dimension (FD) calculated. The FD was evaluated in the time interval corresponding to the motor imagery and averaged out for all the subjects (each channel). In order to characterize the FD distribution, the linear regression curves of FD over the electrodes position were applied. The differences FD between the proposed mental tasks are quanti ed and evaluated for each experimental subject. The obtained results of the proposed method are a substantial fractal dimension in the EEG signal of motor imagery tasks and can be considerably utilized as the multiple-states BCI applications. Keywords— electroencephalogram (EEG), motor imagery tasks, mental tasks, biomedical signals processing, human-machine interface, fractal analysis, critical exponent method (CEM).