Abstracts

Rationale: Loss of hippocampal internal architecture (HIA) is a classic feature of hippocampal sclerosis, but conventional MRI does a poor job of demonstrating HIA consistently. We previously developed a method, HR-MICRA, to do a superior job of showing HIA and compare its performance at both 3T and 7T. Methods: Subjects included 12 controls and 5 patients with temporal lobe epilepsy. Images were acquired on a Siemens 3T Prisma system and Siemens 7T Magnatom system. A T2w 3D SPACE sequence optimized for good gray-white contrast and short time of acquisition of each individual scan, ~6.5 min on each machine. The resolution at 3T was 0.5 mm in the coronal plane with 0.7 mm slices, and 0.35 mm in the coronal plane with 0.5 mm slices at 7T. Each scan was repeated multiple times, coregistered, and averaged. At 3T, 6 scans were averaged, and at 7T the number of scans varied as the protocol evolved. After review of initial 7T scans, mean images of 4 or 6 averages were found to have an undesirable amount of noise, so the protocol was revised to include 9 or 10 scans as time allowed. Images were compared side-by-side for subfield definition and were scored using a HIA rating system (Ver Hoef et al 2014). The head, body and tail were scored separately because the morphology of each is strikingly different resulting in differences in HIA clarity. Results: Five averaged scans were uninterpretable due to movement artifact -- 3 at 3T, 2 at 7T. Twelve subjects had scans free of movement artifact, which allowed a side-by-side comparison of 3T vs. 7T in regard to subjective perception of subfield definition. Of these 2 had better subfield definition at 3T than 7T, but both of those had only 4 7T scans averaged; 10 had better subfield definition at 7T though only in 3 of 10 was 7T markedly better across all hippocampal segments, while in 7 of 10, 7T was considered only slightly better than 3T. Table 1a shows the proportions of subjects who had all or virtually all slices in a given hippocampal segment with an HIA score of 3 or better, separated out by head, body and tail. An HIA score of 3 or better is necessary for direct visualization of all subfields. Table 1b shows the proportions that achieved an HIA score of 4, which is the maximum HIA clarity score. Example images are shown in the figure.In one subject with drug-resistant temporal lobe epilepsy and an atrophic hippocampus, there was a notable asymmetry of internal architecture clarity on the affected side compared to the normal side in the 3T images, but HIA clarity was equivalent in the 7T images, suggesting there may be additional benefit to 7T in epileptic hippocampi when subfields are atrophic. Conclusions: Head-to-head comparisons of 3T and 7T mean images showed that the reviewer thought that hippocampal subfields were more clearly visualized in the 7T images, though in many cases the difference was slight. Using an optimized number of averages (9-10), 7T was superior to 3T, with good or excellent HIA clarity in the body and tail in all subjects, and in the head in the majority of subjects (71%). Across both platforms, imaging HIA clearly was most challenging in the hippocampal head, and somewhat more challenging in the tail than the body. Funding: NINDS R01NS094743