Abstracts

Rationale: Understanding how the hippocampus is connected to the rest of the neocortex has important implications for temporal lobe resection and post-surgical outcomes. Many theories exist about the functional differences between the anterior and posterior hippocampus and most studies have used clear demarcations to measure this. Here we leveraged an advanced resting-state fMRI analytical method called connectopic mapping (Haak et al., 2017) to investigate hippocampal functional organization to the cortex without a specific parcellation scheme. Connectopic mapping was applied to resting-state fMRI data to estimate hippocampal-neocortical functional at the individual level for typically developing (TD) children and patients with focal epilepsy.

Methods: We analyzed 22 patients with focal epilepsy and 633 TD participants from the Human Connectome Project-Development (HCP-D) as a large-sample reference group (ages 6-21). First, connectopic mapping estimated an individual ‘fingerprint’ of hippocampal connectivity. For this, we computed the correlation between voxel-wise time-series and the rest of the neocortex, as well as within-ROI similarity of functional connectivity, and applied non-linear manifold learning (Lapalcian Eigenmaps) to the graph representation of this similarity matrix. This resulted in connectopic maps, which collectively represent how hippocampal-neocortical connections vary topographically along the anterior-posterior hippocampal axis. To interpret the topographic changes at the edge level, we conducted seed-based connectivity analysis (Pearson correlations) from multiple significant areas (from group comparison) along the hippocampal longitudinal axis. We then compared these maps between focal epilepsy and TD groups.

Results: For both TD and epilepsy groups, the primary connectopic map, representing the first dominant mode of connectivity change, followed the expected anterior-to-posterior trajectory (Figures 1A and 1C). However, the projection maps differed between the groups. Specifically, for TDs the anterior-to-posterior gradient in the projection map represents a gradual connectivity change from regions associated with language (superior temporal gyrus/temporal pole, Broca’s area, BA39/40; Figure 1B) to those subserving visual and spatial processing (visual cortex, superior parietal lobule) and executive functioning. For the epilepsy group, however, the projection map did not follow the same anterior-to-posterior trend (Figure 1D).

Conclusions: Similar to Haak et al. (2017) with TD adults, we found that functional organization of the hippocampal longitudinal axis is gradual in TD children and those with epilepsy, and not segregated into parcels. This anterior-to-posterior gradient projects to expected language (for anterior hippocampus) and visual/executive functioning (for posterior hippocampus) regions for TD children; however, this is disrupted in pediatric epilepsy.

Funding: NINDS K Award 1K23NS093152-01