Abstracts

Rationale: Epilepsy is a network disorder which affects how brain regions communicate. Network communication, or connectivity, is investigated through magnetic resonance imaging (MRI). Structural connectivity (SC) is examined through diffusion MRI (dMRI), which estimates how many physical connections exist between regions. Functional connectivity (FC) is examined through resting-state functional MRI (rs-fMRI) and uses the correlation of haemodynamic activity to estimate how regions communicate regardless of their anatomical connections. Structure-function coupling (SFC) examines the interplay between SC and FC, and has the ability to uncover additional information compared to either modality individually, but has not been studied in pediatric focal epilepsy. In this preliminary work, we investigate the ability of SFC to identify network alterations compared to SC and FC individually.

Methods: We plan to enroll 50 patients (aged 7-18) within 5 years of focal epilepsy diagnosis. Healthy control data (n=348) was age matched from the Lifespan Human Connectome Project-Development (HCP-D) study. Participants were scanned with a protocol (matched to HCP-D) on a 3 Tesla Siemens Prisma MRI with a 32-channel head coil including high-resolution 0.8 mm T1w and T2w images with prospective motion correction, 2 mm rs-fMRI, and 1.5 mm multi-shell dMRI sequences. rs-fMRI was preprocessed, denoised, and warped to a standard Montreal Neurological Institute template. dMRI was denoised, corrected for artefacts, susceptibility, motion and distortions. Probabilistic tractography was derived from dMRI within a white matter mask. The Schaefer parcellation (300 regions) was applied to identify regional SC and FC, performing Spearman correlations on corresponding columns of respective connectivity matrices to calculate SFC. SC, FC and SFC were averaged for each of the Yeo-7 networks (visual, default, somatomotor, dorsal attention, ventral attention, limbic and frontoparietal control) to obtain network-specific SC, FC and SFC (Figure 1).

Results: Preliminary results are presented on the first six subjects recruited. SC, FC and SFC of patients were normalized to controls in each network (Figure 2). Independent t-tests were used to compare average SC, FC and SFC of patients to controls in each network. Significant differences (p< 0.05, uncorrected) were identified in the left visual SFC and right limbic SFC networks. No significant differences were identified in SC or FC.