Modeling and simulation of EEG signals for diagnosis of epilepsy

ABSTRACT

AR [autoregressive modeling] and ARMA [autoregressive moving average modeling] are very popular in modeling processes, in the present work they were used to simulate the EEG [brain activity] to investigate which of them is better for simulating the brain activity. Two groups were investigated; control group and Epileptic group [patients suffer from brain disorder and convulsions], it was found that AR model is efficient to simulate the brain activity, because of its stability and its coefficients of variation fluctuate in a limited range. ARMA model showed higher values for the calculated variation coefficient of model parameter. The coefficient of variation was reached 182% for 80% of the model coefficients. This variation was obvious over the studied time intervals. In addition, ARMA model needs more time of computation, therefore AR model is preferred to simulate the brain activity