Purpose or Case Report: Purpose: The evaluation of epileptic foci with FDG PET can be challenging, particularly when epileptic foci are subtle. We sought to determine if an age-matched database of normative cerebral FDG PET uptake can be used to automatically identify epileptic foci.
Methods & Materials: Materials and Methods: Through an IRB-approved study, we constructed an in-house normative database for FDG cerebral uptake using the brain portion of total Body PET scans (n=84) performed for oncology assessment with no known neurologic pathology. We separated the data into age bins (0-2, 3-5, 6-8, 9-11, 12-14, 15-17 years), and nonlinearly co-registered the studies to an age-matched MRI atlas by age (18 mo, 4 yr, 7 yr, 10 yr, 13 yr, 16 yr) using BioImage Suite. Within each age bin, each registered brain image was normalized such that the mean of the PET voxels within the MRI mask was the same across patients. The mean and standard deviation for each age bin formed the normative atlas. This atlas was applied to PET brain studies performed for epilepsy assessment (n=10). Assessment of the epilepsy studies followed the same processing as entries for the atlas (registration, normalization within each age bin), and were then compared to the normative atlas in terms of ratio from mean and number of standard deviations from mean (z-scoring). The detection performance was evaluating by comparing the automatically identified epileptic zones to the original radiologist report.
Results: Results: The FDG PET atlas was generated with 84 normal studies and tested with 10 abnormal epilepsy exams (age 2-16 yrs). The radiologist reports identified 15 epileptic foci in the abnormal studies. Using regions exceeding 4 standard deviations as a threshold for epileptic foci, the automatic identification with the normative atlas identified 14 of these sites (93% Sensitivity) and also identified 7 additional false positive sites.
Conclusions: Conclusions: An age-matched normative FDG PET Brain database can provide a highly sensitive tool to help identify FDG PET abnormalities in epilepsy exams. Additional refinements are needed to improve the specificity of this approach.