Recent publications on the statistics of mental disorders have reported that schizophrenia (SZ) affects approximately 20 million people worldwide. It can be characterized by array of symptoms including delusions, hallucinations, disorganized speech, decrease in emotions, loss of enjoyment, social withdrawal, or difficulty in understanding and learning. The precise cause of SZ is still under investigations: however, it has been understood that abnormalities in neurotransmissions, mainly abnormal activity or block in dopamine receptors could be one of the pathophysiological conditions. From the electrophysiological point of view the electromagnetic activity in schizophrenia is often analyzed using 16 main regions of interests in each hemisphere with 4 in each lobe. In this paper we develop an inverse model that estimates probability density function of the dipole sources using maximum likelihood estimation. To account for model inaccuracy, we will use different model to estimate the unknown activity of the cortical surfaces based on the Gaussian mixture model. To evaluate the applicability of the proposed technique we simulate EEG measurements using cortex region activity parameters reported in clinical studies and add measurement noise to account for the noise in actual EEG systems. Finally, we compare the performance of our model using the goodness-of-fit with respect to the real EEG/MRI measurements acquired in the clinical studies.