Persistent white matter disruption underlies apparent functional normalization in intractable temporal lobe epilepsy: Evidence from multimodal MRI.
Authors
Affiliations (3)
Affiliations (3)
- Department of Neurology, People's Hospital of Ningxia Hui Autonomous Region, Ningxia Medical University, Yinchuan 750002, China.
- Department of Neurology, The First Affiliated Hospital of Guangxi Medical University, Nanning 530021, China.
- Department of Neurology, Ruikang Hospital Affiliated to Guangxi University of Chinese Medicine, Nanning 530011, China.
Abstract
Intractable temporal lobe epilepsy (ITLE) poses ongoing therapeutic challenges due to resistance to antiseizure medications and limited improvements in patient quality of life. Although neuroimaging studies have identified both functional and structural abnormalities in ITLE, their combined analysis and relevance to treatment refractoriness remain unclear. We collected longitudinal rs-fMRI and DTI data from 29 patients with ITLE (pre-/post-treatment) and 25 healthy controls and extracted ALFF/ReHo and FA/MD/AD/RD features. Primary inferences were based on conventional group and longitudinal analyses, while exploratory random-forest classification and UMAP visualization were used to compare functional and structural discriminative patterns. Post-treatment functional measures in ITLE patients showed partial normalization toward healthy control levels, whereas white matter abnormalities-especially reduced fractional anisotropy in the Fornix (crus)/Stria terminalis-remained persistent. In exploratory classification analyses, structural features continued to distinguish ITLE from controls after treatment, while functional features more frequently classified post-treatment ITLE cases as healthy, consistent with a structural-functional dissociation. These findings suggest that persistent white matter microstructural alterations may contribute to treatment resistance in ITLE despite apparent functional normalization. Exploratory machine-learning and low-dimensional visualization offered converging, quantitative support for this dissociation and may help prioritize candidate targets for future validation in larger cohorts.