Authors :
Presenting Author: Benjamin Speidel, MS – University of California, San Francisco
Jonathan Kleen, MD, PhD – University of California, San Francisco
Robert Knowlton, MD, MSPH – University of California, San Francisco
Rationale:
Localization of seizure foci in magnetic resonance imaging (MRI)-negative refractory epilepsy remains challenging. A new atlas-based quantitative tool computes standardized uptake value ratio (SUVR) percentiles and asymmetry indices for each Desikan–Killiany region versus 37 healthy controls to highlight subtle hypometabolism. We tested whether access to this normative comparison improves reader confidence and inter-rater agreement in detecting focal hypometabolism on fluorodeoxyglucose positron emission tomography (FDG-PET).Methods:
Fifty-three MRI-negative refractory epilepsy patients underwent interictal FDG-PET co-registered to MRI. Two independent reviewers visually inspected each scan, marked regions of interest (ROIs) of potential hypometabolism using the Desikan–Killiany atlas, and rated confidence (1–10). They then re-reviewed each case with the quantitative tool, which (1) computes SUVR (normalized to whole-brain uptake), (2) derives an asymmetry index, and (3) maps both measures to percentile ranks against 37 controls—displayed as an SUVR percentile heatmap and summary table of all metrics. Equipped with this perspective, reviewers again annotated ROIs, rated confidence, and noted tool utility.
Results:
Reviewer confidence increased from a mean of 5.93 to 6.87 (p < 0.0001, paired t-test) after using the quantitative PET tool. Regarding localization of hypometabolic areas, ROIs were altered or refined in 32% of cases after incorporating the tool in their interpretations. Comparison between scorer localizations (areas annotated as involved or not) resulted in Cohen’s kappa values of 0.4172 without the tool to 0.4766 with the tool (increase of 0.0594). Reviewers rated the heatmap and tables as helpful in a mean of 78% of cases.Conclusions:
In MRI-negative refractory epilepsy, augmenting visual FDG-PET interpretation with region‐based SUVR percentiles and asymmetry indices significantly improves reviewer confidence, refines hypometabolic focus localization and improves inter-rater agreement. This quantitative tool may reduce the subjectivity in PET interpretation to support more accurate presurgical planning and warrants further validation.
Funding: n/a