Single-subject gray matter networks in temporal lobe epilepsy patients with hippocampal sclerosis

Abstract

ObjectivePrevious studies have demonstrated structural brain network abnormalities in patients with temporal lobe epilepsy (TLE) using cortical thickness or gray matter (GM) volume. However, no studies have applied single-subject GM network analysis. Here, we first applied an analysis of similarity-based single-subject GM networks to individual patients with TLE. Materials and methodsWe recruited 51 patients with TLE and unilateral hippocampal sclerosis (22 left, 29 right TLE) and 51 age- and gender- matched healthy controls. Single-subject structural networks were extracted from three-dimensional T1-weighted magnetic resonance images for each subject. In this method, nodes were defined as small cortical regions and edges representing connecting regions that have high statistical similarity. The constructed graphs were analyzed using the graph theoretical approach. The following global and local network properties were calculated: betweenness centrality, clustering coefficient, and characteristic path length. In addition, small world properties (normalized path length λ, normalized clustering coefficient γ, and small-world network value σ) were obtained and compared with those for the controls. ResultsAlthough the small-world configurations were retained, impaired global clustering coefficient was observed in left and right TLE. At a regional level, patients with left TLE showed a widespread decrease of the clustering coefficient beyond the ipsilateral temporal lobe and a decreased characteristic path length in the ipsilateral temporal pole. On the other hand, patients with right TLE showed a localized decrease of the clustering coefficient in the ipsilateral temporal lobe. ConclusionsOur findings suggest that global and local network properties disrupted and moved toward randomized networks in TLE patients in comparison to controls. This network alteration was more extensive in left TLE than in right TLE patients. Single-subject GM networks may contribute to a better understanding of the pathophysiology of TLE.