Abstracts

Rationale: Measurement of hippocampal subfield volumes has been predicted to be useful in temporal lobe epilepsy (TLE), based on the presence of subfield-specific pathology in this condition. Using MRI with high spatial resolution (which is possible at high magnetic field strength), hippocampal subfield volumetry has been performed in a variety of conditions - including epilepsy. We have developed a hippocampal subfield segmentation method using a high resolution fast spin echo (FSE) sequence. In this study, we aimed to evaluate the ability of this technique to delineate the hipppocampal subfields in a cohort of control and TLE subjects. Methods: Twenty-five normal control subjects (age 20-57) and ten patients (age 23-58) with unilateral (five right) TLE were studied. Nine of the patients had qualitative evidence of hippocampal sclerosis (HS) (increased T2 signal and hippocampal atrophy) on 1.5-T MRI unilaterally; one patient had normal MRI and depth electrodes indicating unilateral seizure onset. Five of the patients have undergone surgery (four HS and one focal cortical dysplasia). Subjects were imaged with 4.7-T MRI using an FSE sequence (native resolution: 0.52 X 0.68 X 1.0 mm). A recently published hippocampal internal architecture scoring system [Ver Hoef et al.; Epilepsy Res. (2013) 106, 146-154] was modified to classify hippocampi as either: i) adequate for segmentation: moderate to high contrast between SLM and hippocampal grey matter; or ii) inadequate for segmentation: minimal or no contrast between SLM and hippocampal grey matter. Hippocampi were rated bilaterally in each subject by a single observer, who was blinded to subject category (patient versus control). Presence or absence of motion artifacts was assessed, in order to exclude this as a potential confounding variable. Results: The frequency of significant motion artifacts was not significantly different between the two groups (p=0.36), being identified in 16% (4/25) of controls, and 30% (3/10) of TLE patients. Inadequate visualization of the SLM was significantly (p=0.0001) more common in the TLE group hippocampi (17/20, 85%) compared with control group hippocampi (9/50, 18%). In all ten patients with TLE, loss of internal architecture was identified in the symptomatic hippocampus - with seven patients having impaired visualization of the SLM bilaterally. Conclusions: Our results demonstrate that the majority of patients with TLE have bilateral abnormalities of hippocampal architecture (inability to adequately visualize the SLM) on high resolution T2-weighted images. This finding is significant as it suggests that while high spatial resolution FSE imaging is sufficient for hippocampal subfield segmentation in nonepileptic subjects, it may be inadequate for hippocampal subfield segmentation in TLE. Novel MRI acquisition and post-processing methods must be explored to allow more accurate hippocampal subfield segmentation in this condition.