Localization of seizure sources using blind identification and a new clustering algorithm

In this paper, a new method is proposed to localize the seizure sources from multi-channel electroencephalogram (EEG) signals. The second order blind identification (SOBI) is applied to estimate the brain source signals in a number of signal segments. After calculating the unmixing matrices in several time frames, the rows of these matrices are clustered by a new proposed clustering method. By multiplying each cluster center to the electrode signals (EEGs), the brain signal sources are approximated. According to short term largest Lyapunov exponent values, the main seizure source signal is separated from the others. This source signal is projected back to the electrodespsila space in order to localize the seizure source in the brain. The simulation results and also the clinical tests derived using the simultaneous intracranial recordings verify the accuracy of the system.