Utilising MRI Data and VOI-Based Analysis for Enhanced Epilepsy Prediction: A Translational Approach from Bench to Bedside

Abstract

Background: Epilepsy is one of the most prevalent neurological disorders globally, profoundly impacting patient’s quality of life and stretching healthcare resources. Despite technological advances in neuroimaging, early and accurate detection of epileptic foci remains elusive, especially when standard MRI scans appear structurally normal. For clinicians, radiologists and neurologists, the limitations of subjective interpretation underscore an urgent need for objective diagnostic methods. This study addresses that gap by introducing Volume of Interest (VOI)-based analysis as an innovative tool to detect microstructural brain abnormalities associated with epilepsy. Methods: This study involves a retrospective analysis of 30 brain MRI images (T1-Weighted MPRAGE), comprising of 12 epilepsy patients and 18 matched controls (11 males, 19 females; mean age 42.7 ± 17.2 years). The images were converted from the DICOM format into the Brain Imaging Data Structures (BIDS) standard using the fMRIPrep platform, which were then normalised following the standard Montreal Neurological Institute (MNI) template. The images were segmented into Grey Matter, White Matter and Cerebrospinal Fluid. The segmented Grey Matter regions have then been analysed using VOI-based analysis from the Harvard-Oxford Cortical Structural Atlas. The VOI-based analysis results were then statistically tested using ANCOVA with False Detection Rate (FDR) correction. Results: Three regions show statistically significant ANCOVA results, which were Superior Frontal Gyrus (p=0.041), Superior Parietal Lobule (p=0.026) and Lingual Gyrus (p=0.036). However, all three regions fail the FDR correction (q=0.36). Conclusion: This work shows that combining MRI data with VOI-based analysis can reveal subtle structural patterns in the superior frontal gyrus, superior parietal lobule and lingual gyrus that may contribute to the understanding of epilepsy risk. Although these patterns were not sustained after FDR correction, they offer a promising direction for more objective predictive tools. Continued research with larger cohorts is essential to confirm these early signals and strengthen their relevance for epilepsy prediction.