GETool software tool for geometric localization of cortical abnormalities in epileptic children

This thesis presents GETool, a new software that allows objective localization of cortical abnormalities in children who are being considered for epilepsy surgery. Muzik and Draghici have proposed the geometric parcellation approach to detect epileptic foci. We present a novel method of geometric parcellation for the non-geometric cortical surface. The method involves performing parcellation on a sphere which is then mapped onto the brain surface. This gives us standardized, homotopic, non-overlapping cortical regions. The PET concentration in these cortical regions is calculated for a group of normal adults and the patient using a variant of the normal fusion approach. Histograms are obtained which characterize the tracer distribution at specific cortical locations. We determine the difference area between the patient and the mean histogram of the group of normal adults. The functional abnormality of a cortical region of the patient at a given location and resolution level can be detected based on the difference area. We present the validation results of geometric parcellation using the gold standard of subdural EEG. GETool provides objective definition of the size and location of unilateral and bilateral cortical abnormalities and provides guidance in the placement of subdural EEG grids.