Mendelian randomization and clinical retrospective radiomics study reveal causal relationships between brain imaging-derived phenotypes and epilepsy risk.

This study investigates the causal relationships between brain imaging-derived phenotypes (IDPs) and epilepsy risk by integrating Mendelian randomization (MR) analysis and radiomics-based validation using brain magnetic resonance imaging (MRI) data. Two-sample MR analysis, leveraging genome-wide association study (GWAS) data, revealed that higher fractional anisotropy (FA) in the right posterior thalamic radiation significantly increases epilepsy risk (HR = 1.31, 95% CI: 1.16–1.49), while larger volumes of the right pallidum exert a protective effect (HR = 0.67, 95% CI: 0.55–0.80). Radiomics analysis of T1-weighted MRI images identified key structural features in these brain regions, which were incorporated into machine learning models for robust epilepsy risk prediction, confirming the MR findings. These results provide novel insights into the structural mechanisms underlying epilepsy, emphasizing the right posterior thalamic radiation and right pallidum as key regions influencing risk. By integrating MR and radiomics approaches, this study highlights the potential of imaging-genetics methodologies in advancing our understanding of epilepsy and improving clinical decision-making.