Manifold-Derived Gradients Link the Cortical Morphology to Gene Expression in Patients with Epilepsy

Abstract

Diffusion embedding techniques offer the means to map brain organizational principles into a single manifold, called gradients, which help us understand how multi-scale brain features are related to each other. These links can help us understand the neural mechanism, and further broaden our insight into the relation between structure and function. In this study, manifold gradient was leveraged to study the link of macroscopic morphology to cortical gene expression in 92 genetic generalized epilepsy with generalized tonic-clonic seizures (GTCS), and 116 healthy controls. Diffusion map embedding was used to estimate the gradients of morphometric similarity (MS) connectome, which described the interregional similarity through five cortical features. The results revealed significant between-group differences in the MS gradient in GTCS, mainly involved in high-level networks. And the changes in MS gradient showed a positively correlation between expression profile of epileptic risk genes.