Objective detection of epileptic foci by 18F-FDG PET in children undergoing epilepsy surgery.

UNLABELLED PET has been used for the presurgical localization of epileptic foci for more than 20 y; still, its clinical role in children with intractable epilepsy remains unclear, largely because of variable analytic approaches and different outcome measures. The purpose of the present study was to evaluate and optimize the performance (lateralization and lobar localization value of epileptic foci) of objective voxel-based analysis of (18)F-FDG PET scans in a pediatric epilepsy population. METHODS Twenty children with intractable focal epilepsy (mean age ± SD, 11 ± 4 y; age range, 6-18 y) who underwent interictal (18)F-FDG PET, followed by 2-stage epilepsy surgery with chronic subdural electrocorticographic monitoring, and were seizure-free after surgery were included in this study. PET images were analyzed using both a visual-analysis and a statistical parametric mapping (SPM) method. Lateralization value and performance of lobar localization (in lateral and medial surfaces of all lobes, total of 8 regions in each epileptic hemisphere), calculated for 3 different statistical thresholds, were determined against intracranial electrocorticography-determined seizure-onset region and surgical resection site. RESULTS SPM using a statistical threshold of P less than 0.001 provided 100% correct lateralization, which was better than visual assessment (90%). Although visual and SPM analyses (with both P < 0.001 and P < 0.0001 thresholds) performed similarly well (with a sensitivity and specificity of 74% or above) in the localization of seizure-onset regions, SPM detected 7 of 9 seizure-onset regions, mostly in medial cortices, that were missed by visual assessment. Also, SPM performed equally well in both hemispheres, compared with visual analysis, which performed better in the left hemisphere. No statistical difference in performance was observed between visual and SPM analyses of children with abnormal versus normal MRI findings or of children with gliosis versus developmental pathology. Clinical variables, such as age, duration of epilepsy, age of seizure onset, and time between PET and last seizure, showed no correlation with sensitivity or specificity of either visual analysis or SPM analysis. CONCLUSION SPM analysis, using a young adult control group, can be used as a complementary objective analytic method in identifying epileptogenic lobar regions by (18)F-FDG PET in children older than 6 y.