Abstracts

Rationale: High resolution MRI has revolutionized the identification and surgical treatment of patients with mesial temporal lobe epilepsy (mTLE). Hippocampal sclerosis (HS) on MRI usually presents with atrophy of the hippocampus and hyper-intensity in FLAIR. However another group of temporal lobe epilepsy patients exists, whose HS are so subtle that the MRIs are completely negative (MRI-). In these cases, an objective, quantitative MRI post-processing technique that could help lateralize the pathological hippocampus would be very desirable. The aim of our study is to apply a recently published quantitative hippocampal FLAIR analysis in a strictly defined cohort of MRI-negative, pathologically verified mTLE patients. Methods: From the surgical series of Cleveland Clinic Epilepsy Center between 2002.1 and 2011.12, patients were included if they: 1) underwent anterior temporal lobe resection with mesial structures; 2) were considered MRI- prior to surgery, and (3) HS pathologically verified. All MRI scans had been evaluated prospectively by dedicated epilepsy neuroradiologists at the time of pre-surgical evaluation, and evaluated again with epileptologists during patient management conference, where results of all noninvasive investigations were considered. For a strictly defined MRI- cohort, we only selected patients who were still considered MRI- after the two-step reviewing process. Nine patients fulfilled all requirements and were analyzed using a novel MRI post-processing technique for automated quantitative hippocampal FLAIR analysis based on algorithms of SPM5 (Huppertz et al., Epilepsy Res. 2011). The T2 FLAIR images underwent spatial and intensity normalization, and were thresholded by a probabilistic hippocampal mask. The resulting left and right hippocampal images were then calculated for the mean voxel intensity and compared to each other. This analysis was performed on upper quartile of voxels belonging to each hippocampus. Lateralization from the analysis was then compared to histopathology findings by careful re-review of the surgical pathology slides. Results: Out of the 9 patients analyzed, 4 had right-sided HS and 5 had left-sided HS according to histopathology. One patient had co-existing FCD Ia in the neocortical temporal structure. Actual value of the average hippocampal intensity can be quite variable among the patients. However, in all 9 patients, the side of hippocampus with higher average intensity correlates well with pathology confirmation (Figure 1). Conclusions: Improving noninvasive lateralization of MRI-negative mTLE cases is critical to management of the epilepsy. Our experience using the automated quantitative FLAIR analysis in a strictly defined MRI-negative cohort supports that this technique can be a valuable additional tool to the presurgical evaluation of suspected mTLE patients with negative or questionable MRI findings. Further refinement of this technique would involve generation of a normative scanner-specific database so that descriptive statistics can be calculated.