Structural connectivity changes in temporal lobe epilepsy: Spatial features contribute more than topological measures

• Using MRI data, we analyse 22 patients with left TLE and 39 non-epileptic controls.
• With a connectomics approach we investigate how nodal properties such as surface area influence connectivity weight.
• We find significant atrophy (reduced node size) in many brain areas in patients with TLE.
• We show only subtle changes in connectivity.
• When both node size and node connectivity are combined we find significant changes in connection weight.

BackgroundPrevious studies reported reduced volumes of many brain regions for temporal lobe epilepsy (TLE). It has also been suggested that there may be widespread changes in network features of TLE patients. It is not fully understood, however, how these two observations are related.MethodsUsing magnetic resonance imaging data, we perform parcellation of the brains of 22 patients with left TLE and 39 non-epileptic controls. In each parcellated region of interest (ROI) we computed the surface area and, using diffusion tensor imaging and deterministic tractography, infer the number of streamlines and their average length between each pair of connected ROIs. For comparison to previous studies, we use a connectivity ‘weight’ and investigate how ROI surface area, number of streamlines & mean streamline length contribute to such weight.ResultsWe find that although there are widespread significant changes in surface area and position of ROIs in patients compared to controls, the changes in connectivity are much more subtle. Significant changes in connectivity weight can be accounted for by decreased surface area and increased streamline count.ConclusionChanges in the surface area of ROIs can be a reliable biomarker for TLE with a large influence on connectivity. However, changes in structural connectivity via white matter streamlines are more subtle with a relatively lower influence on connection weights.