Abstracts

Rationale: Epilepsy is considered as a disorder of brain networks, with interictal and ictal activity generated and spread across networks that involve one or both hemispheres. Brain networks can be studied with resting-state functional connectivity (FC) metrics. Previous investigations used functional magnetic resonance imaging (MRI), which offers an indirect measure of neural activity, or electrocorticography that has limited spatial resolution and can be used only in patients undergoing neurosurgery. Whole-brain FC measures via electric and magnetic source imaging (ESI/MSI) are powerful tools for noninvasively mapping the neural activity of functionally connected brain regions. Here, we aim to examine global FC in the brain of children with drug resistant epilepsy (DRE). We hypothesize increased FC in the pathological hemisphere of children with DRE compared to their contralateral hemisphere and both hemispheres of typically developing (TD) children.

Methods: We prospectively analyzed simultaneous high-density electroencephalography (HD-EEG) and magnetoencephalography (MEG) data from 9 children with DRE (4 females, age: 12.9 years, median: 14 years) and 9 TD children (6 females, age: 10.2 years, median: 10 years) (Figure 1A). We first constructed a realistic head model from each child’s MRI, and defined regions of interest (ROIs) through volume parcellation based on the automated anatomical labelling atlas. We then performed combined ESI/MSI to reconstruct each ROI’s neuronal activity in the whole-brain, as well as separately for the left and right hemispheres (Figure 1B). Data covariance was computed on 3-minute duration artifact-free segments. For each ROI, we estimated the Amplitude Envelope Correlation (AEC) and corrected imaginary Phase Locking Value (ciPLV) for physiologically relevant frequency bands: delta, theta, alpha, beta, gamma, and broadband (Figure 1B). We used these FC matrices to compute betweenness and closeness centrality metrics of the Minimum Spanning Tree (Figure 1B). We finally estimated a unique mean whole-brain FC value for each child and one for each hemisphere. We compared global FC measures between children with DRE and TD children (Wilcoxon rank-sum test). We also compared FC in the pathological (vs. the contralateral) hemisphere for children with DRE (Wilcoxon signed-rank test).

Results: We found higher global FC in patients with DRE compared to TD for the: (1) ciPLV in gamma band (p=0.01); and (2) closeness centrality of AEC in gamma (p=0.03) and broad bands (p=0.02) (Figure 2A). For patients with DRE, we found higher ciPLV in beta band (p=0.03) inside the pathological hemisphere, as well as higher betweenness centrality for ciPLV in beta band (p=0.03) (Figure 2B).

Conclusions: Building on recent advances of neuroimaging and signal processing, we present here evidence of aberrant FC in children with DRE that are more prominent in the hemisphere of epileptogenic focus. Our findings may provide a better understanding of brain networks and potentially help guide surgical planning in patients with DRE.

Funding: RO1NS104116-01A1 and R21NS101373-01A1 by NINDS