To investigate the utility of quantitative fluorodeoxyglucose-positron emission tomography (FDG-PET) analysis for seizure-onset lateralization with MRI-guided segmentation.

Identifying the seizure focus is a challenging step in surgical evaluation of pharmacoresistant epilepsy. FDG-PET is a powerful tool for identifying focal and regional metabolic changes, but is clinically limited to qualitative interpretation of images. However, extracting quantitative FDG uptake values may improve lateralization and localization of seizure foci.

Patients with FDG-PET and volumetric MRIs were identified from the GWU Epilepsy Surgery Database. MRIs were reconstructed using Freesurfer (v6.0), and PET images were co-registered, pre-processed, and segmented with MRI using PETsurfer. Standardized parcellation atlases were used based on surface anatomy or functional MRI in healthy adults, allowing for comparison between homologous cortices in the native brain space and laterality index calculation for cortical/sub-cortical segments. Seizure focus lateralization was determined from non-invasive EEG. FDG uptake values relative to the pons and non-rescaled values were calculated from co-registered PET/MRIs. The two groups were compared using Student’s t-test or Mann-Whitney U test, and surface-based general linear model analysis of the voxels and cortical segments.

28 patients were included. Using functional segmentation methods, we detected network abnormalities of FDG uptake ipsilateral to the seizure focus in the limbic system, default mode network, and ventral attention system (all p<.05). Anatomical segmentation of cortical and subcortical values had significant lateralizing values in the hippocampus (p<.01), left/right pallidum (p<.05), and the temporal lobe and entorhinal cortex (p<.05).

Quantification of FDG-PET values based on MRI-guided segmentation provides lateralization and potentially localization values for pre-surgical evaluation. Certain anatomical and functional segments show larger differences between left and right-onset focal epilepsies; together these may contribute to higher FDG-PET sensitivity and specificity. Further studies using statistical and machine learning models are underway to establish lateralizing and localizing values of quantitative PET in clinical settings.