Abstracts

Rationale: High resolution MRI has revolutionized surgical treatment of patients with mesial temporal lobe epilepsy (MTLE). Hippocampal sclerosis (HS) on MRI usually presents with hyper-intensity in FLAIR. The objective of this study is to test the effectiveness of a MRI post-processing technique that could help quantify FLAIR signal change in the hippocampus. We chose to study a cohort of patients with MTLE, and a control cohort of patients with neocortical epilepsy. Methods: Patients were retrospectively identified from the surgical series of Cleveland Clinic Epilepsy Center. We included 9 patients who underwent temporal lobe resection with mesial structures, and became seizure-free at one-year followup. For the control group, we included 11 patients who underwent neocortical resections without removing the mesial structures, and became seizure-free at one-year followup. All patients included had high-resolution 3T FLAIR images and surgical pathology. Patients were excluded if they had structure abnormalities near the mesial structures, or if the image quality was low due to motion artifact. We performed chart review to collect the conclusion from the initial MRI report used during the patient's pre-surgical workup. Visual re-analysis was additionally performed by a dedicated neuroradiologist. Automated whole-brain FLAIR analysis was performed on a voxel basis, following methods from Huppertz et al in MATLAB SPM12. Both visual re-analysis and quantitative MRI analysis were performed blinded to the patients' clinical information. Findings from the initial MRI report, visual re-analysis and quantitative analysis were dichotomized and compared, and correlated with histopathology. Sensitivity and specificity were calculated. Results: In the MTLE group, all patients had definite HS ILAE Type I by pathology. Eight of these 9 patients were judged to be abnormal by visual re-analysis or quantitative MRI analysis. The consistency of these two analyses was 78%. Seven of these 9 patients were abnormal according to the initial MRI report. The consistency between the initial MRI report and visual re-analysis was 57%. All 11 patients in the neocortical epilepsy group were judged to be normal by visual re-analysis and quantitative MRI analysis, but two patients were read as abnormal by the initial MRI report. Using the existence of HS as the gold standard, the sensitivity and specificity of visual re-analysis and quantitative analyses were the same, 89% and 100%, respectively. However, the sensitivity and specificity of initial MRI report were lower (78% and 82%, respectively). Figure 1 showcases four patients whose quantitative MRI and visual re-analysis showed consistent and correct results. Conclusions: Quantitative whole-brain FLAIR analysis on hippocampus can be successfully applied in patients with MTLE to identify hippocampal signal change. Both visual re-analysis by experienced neuroradiologist and quantitative methods showed high sensitivity and specificity. The initial MRI report had lowest sensitivity and specificity, and may need to be re-analyzed by dedicated neuroradiologists, especially for epilepsy surgical candidates. Funding: None