MRI-quality and morphometric MRI analysis to identify focal cortical dysplasia: An exploratory study

Abstract

Background: In the pre-surgical evaluation of people with focal epilepsy and a normal MRI, Morphometric Analysis Program v2018 (MAP18) aids in detecting visually inconspicuous focal cortical dysplasia (FCD). We investigated the impact of MRI scans with reduced signal-to-noise ratio (SNR) and spatial resolution (SR) on FCD detection by MAP18, aiming to improve the chances of achieving seizure freedom through epilepsy surgery. Methods: Thirty MRI scans with the identified lesion using MAP18 radiologically confirmed as FCD by a neuroradiologist, were retrospective analysed. SNR and SR were artificially reduced in ten steps, and their impact on MAP18 outcomes was assessed using multilevel analysis. Results: There was a significant effect after reducing SR and SNR for z-score and volume of the FCD cluster, the total number of detected clusters, and volume of these clusters. After SNR reduction, there was also a significant effect for z-score of the total number of detected clusters. FCD became undetectable by MAP18 after six steps of SR reduction (voxel size 2.8 × 2.8 × 2.8 mm³) and after two steps of SNR reduction. Conclusions: This exploratory study suggests that reduced SR and SNR negatively affect FCD detection with MRI post-processing (MAP18). The MAP18 evaluator should screen MRI quality before post-processing, particularly for scans with significant visual noise or voxel sizes of 2.8 × 2.8 × 2.8 mm³ and upwards, as repeating a low-quality MRI scan is less burdensome than the adverse effects of continued seizures due to failure to detect FCD.