Quantitative [18F]-fluorodeoxyglucose PET analysis with a Gaussian normative template

OBJECTIVE: To develop a standardized quantitative [18F]-fluorodeoxyglucose (FDG) PET normative template and evaluate its clinical utility for objective epileptogenic focus localization in drug-resistant epilepsy. METHODS: [18F]-FDG PET data from 89 participants across three multicenter datasets were harmonized and categorized into: a construction group (n = 55 healthy subjects), an independent specificity validation group (n = 10 healthy subjects), and a clinical detection group (n = 24 temporal lobe epilepsy [TLE] patients). After spatial and intensity normalization, four candidate models (Gaussian, Gamma, Smoothed Gaussian, and Smoothed Gamma) were constructed. Normality was validated via voxel-wise pass rates to select the optimal template, with clinical utility assessed using a z-score threshold of less than -3.5. RESULTS: The Smoothed Gaussian model exhibited the highest voxel pass rate and was established as the final template. Validation in the 10 independent healthy subjects yielded no false positives, confirming high specificity. Furthermore, the template accurately identified metabolic abnormalities in all 24 TLE patients, consistent with clinically confirmed foci. CONCLUSION: The Smoothed Gaussian-derived template enables objective and reproducible [18F]-FDG PET analysis. It significantly improves focus localization accuracy and provides a reliable quantitative tool for guiding drug-resistant epilepsy surgery.