Abstracts

Rationale: We aimed to investigate whether the computerized processing of three-dimensional T1-weighted magnetic resonance imaging (MRI) to enhance the characteristics of focal cortical dysplasia (FCD) was helpful in the detection of possible epileptogenic lesions in epileptic patients.Methods: Forty-seven patients who underwent epilepsy-dedicated 3T-MRI were included. Their age ranged from 2 to 40 years (median age: 13 years). Three-dimensional T1-weighted images with a slice thickness of 1 mm were corrected for inhomogeneities and processed by a method described by Bernasconi et al. (Ann Neurol 2001;49:770-775) with some modifications. Three types of images were made by freely available programs: 1) a gradient map reflecting the clarity of the gray-white matter junction, 2) a thickness map showing the cortical thickness, and 3) a relative intensity map enhancing the signal intensity of the gray matter. Finally a ratio map, in which typical FCD would appear bright, was made by the following formula: thickness map intensity map / gradient map. Two medical students with little MRI experience who were blinded to the clinical characteristics and MRI findings of the patients made joint judgments whether there were signal differences between hemispheres and where they were. The results were compared with MRI lesions identified by a neuroradiologist and an epileptologist using all available neuroimages, electroencephalographic findings, and patients' clinical profiles.Results: There were 11 patients with focal MRI lesions: seven with FCD, two with hippocampal sclerosis, one with focal polymicrogyria, and one with a thalamic lesion. The sensitivity of processed MRI was 81.8% (nine of 11 patients) and specificity was 72.2% (27 of 36 patients). In two patients whose MRI lesions were missed by processed MRI, one had focal polymicrogyria and the other had FCD unclear on T1-weighted images. False-positive detection occurred in 10 patients, with five around the insula and basal ganglia, two around the Sylvian fissure, one in the frontal lobe, one in the temporal lobe, and one with suspicious FCD that was not clearly determined.Conclusions: This analysis will help both community hospitals where experienced neuroradiologists may not be readily available and larger epilepsy centers through the time-efficient identification of focal lesions.