Abstracts

Rationale: Temporal Lobe Epilepsy (TLE) is a common neurological disorder affecting the hippocampus. Around 30% of TLE patients do not respond to medical therapy and are candidates for surgical resection of the epileptic zone. Accurate seizure localization is crucial prior to resection in order to maximize chances of seizure freedom and minimize post-surgical memory deficits. While some prior studies have utilized MRI to lateralize seizure onset zone in TLE patients, limited work has been done investigating structural and functional network architecture of the hippocampus and surrounding medial temporal lobe (MTL) structures at the subregion level. Methods: In this study, we utilized high-resolution 7T resting BOLD-fMRI and MTL-focused 7T T2-MRI to characterize both the functional and structural network properties of the MTL network in medically refractory TLE patients (n=18) compared with healthy controls (n=15). We employed a multi-atlas segmentation algorithm to identify ten MTL subregions per hemisphere (including hippocampal subfields and discrete regions of the parahippocampal gyrus). Structural connectivity matrices were generated from covariance of subregion volumes across subjects, and functional connectivity matrices were generated from linear correlations between fMRI time series in each subregion. In order to assess variability of the data in a common framework for both functional and structural networks, a bootstrapping procedure was carried out separately for functional and structural networks to generate a set of 1000 matrices for the control, lesional TLE, and nonlesional TLE groups. For each set of matrices, global clustering coefficient and efficiency were computed and normalized to connectivity strength. Additionally, functional-structural correlation was determined for each subject group by correlating mean functional and mean structural connectivity strength between each pair of subregions. Results: Compared with controls, we find increased structural clustering, as well as decreased structural efficiency and increased functional efficiency in TLE patients. Effects are more prominent in lesional TLE than in nonlesional TLE (Figure 1). We also find a marked decrease in functional-structural correlation in patients with TLE (r=0.57 for controls, r=0.31 for both lesional and nonlesional TLE, Figure 2).Other than the clustering coefficient for functional networks, all comparisons are significant (p < 0.005) via permutation-based tests. Conclusions: Our findings can help elucidate underlying brain network disruptions in TLE patients who are candidates for surgical resection, including nonlesional patients in which conventional imaging is insufficient. Moreover, this study represents the first comprehensive characterization of functional and structural network connectivity within medial temporal lobe subregions and has important implications for our understanding of MTL topology in both normal and pathological human brains. Funding: NIH Grant T32EB009384, 1 R03 EB16923-01A1, TAPITMAT-TBIC #10037893, CBICA seed grant