Graph theory application with functional connectivity to distinguish left from right temporal lobe epilepsy.

Objective: To investigate the application of graph theory with functional connectivity to distinguish left from right temporal lobe epilepsy (TLE).

Methods: Alterations in functional connectivity within several brain networks - default mode (DMN), attention (AN), limbic (LN), sensorimotor (SMN) and visual (VN) - were examined using resting-state functional MRI (rs-fMRI). The study accrued 21 left and 14 right TLE as well as 17 nonepileptic control subjects. The local nodal degree, a feature of graph theory, was calculated foreach of the brain networks. Multivariate logistic regression analysis was performed to determine the accuracy of identifying seizure laterality based on significant differences in local nodal degree in the selected networks.

Results: Left and right TLE patients showed dissimilar patterns of alteration in functional connectivity when compared to control subjects. Compared with right TLE, patients with left TLE exhibited greater nodal degree' (i.e. hyperconnectivity) with right superomedial frontal gyrus (in DMN), inferior frontal gyrus pars triangularis (in AN), right caudate and left superior temporal gyrus (in LN) and left paracentral lobule (in SMN), while showing lesser nodal degree (i.e. hypoconnectivity) with left temporal pole (in DMN), right insula (in LN), left supplementary motor area (in SMN), and left fusiform gyrus (in VN). The LN showed the highest accuracy of 82.9% among all considered networks in determining laterality of the TLE. By combinations of local degree attributes in the DMN, AN, LN, and VN, logistic regression analysis demonstrated an accuracy of 94.3% by comparison.

Conclusion: Our study demonstrates the utility of graph theory application to brain network analysis as a potential biomarker to assist in the determination of TLE laterality and improve the confidence in presurgical decision-making in cases of TLE.