Abstracts

Rationale: Temporal lobe epilepsy (TLE) due to hippocampal sclerosis is the most common form of medication refractory epilepsy. Radiologists are often asked to determine whether there is evidence of hippocampal abnormalities on high-resolution MRI. This decision can be made with high confidence when there is clear-cut unilateral hippocampal atrophy, but it can be challenging when atrophy is subtle or bilateral. We propose a decision support tool that uses information from a combination of MRI parameters, obtained from T1-weighted and diffusion tensor imaging (DTI) scans, to provide a quantitative score indicating the likelihood of limbic structural abnormalities in patients with TLE. The purpose of this study is to assess the detection accuracy of the proposed tool.Methods: We used T1-weighted images and fractional anisotropy (FA) parametric maps obtained from DTI scans from 13 consecutive patients with unilateral TLE and 28 controls recruited from the local community. All TLE patients had unilateral seizure onset confirmed by prolonged video-electroencephalography recording. T1-weighted images were submitted to voxel-based morphometry yielding maps of gray matter loss. FA images were submitted to voxel-wise analysis yielding maps of white matter deficits. These images were then used as input to an in-house developed package, which constructed a pattern classifier to identify structural abnormalities.Results: Employing T1-weighted images only, the discriminant scores correctly identified abnormalities in 8 (62%) patients. Discriminant scores from FA images alone correctly classified 11 (85%) patients. The combined scores correctly identified regional abnormalities in 12 patients (92% accuracy) (Figure1).Conclusions: With current advancements in multiparametric MRI technologies and leveraging advanced image analytics, there is growing potential for computational decision support tools to improve the accuracy of the clinical evaluation of structural abnormalities in TLE. Automated tools allow for objective assessment of subtle structural deficits that cannot be readily identified by the human eye.