Generation of an Age-Dependent and Harmonized ¹⁸F-FDG Brain PET Atlas Using a High-Sensitivity Short-Axial FOV PET/CT...

¹⁸F-FDG brain PET atlas is widely used in clinical and neuroscience research. However, variation in PET scanner technology can lead to misinterpretation and quantification errors with mismatched atlases. This study aimed to develop high-resolution, age-stratified FDG brain atlases harmonized across different effective image resolutions (EIRs) to support diverse PET/CT scanners. A total of 140 adults (aged 20-90 years, 72 males) were retrospectively enrolled and divided into seven age groups (n = 20). All underwent ¹⁸F-FDG PET/CT at China-Japan Friendship Hospital (Jan 2022-May 2023) using the Siemens Biograph Vision 600 scanner. PET scans (4 min) were acquired 60 min after ¹⁸F-FDG injection (4.44 MBq/kg). System EIR was measured using a Hoffman phantom, and images were smoothed to emulate 5.0-6.0 mm and 8.0-10.0 mm resolutions. Images were normalized to MNI space (1.0 mm) via ANTs, and subjects with outlier FDG uptake values were excluded. SUVRs were calculated using the whole cerebellum as reference. Age-stratified and EIR-harmonized atlases were created. Gray/white contrast (GWC), age correlations, and SUVR differences were analyzed. After quality control, 97 participants were included for atlas construction. Atlases for both EIRs were generated. GWC was higher at 5.0-6.0 mm than at 8.0-10.0 mm (e.g., 2.73 vs. 1.76 in 20-30 years). SUVRs in cortical and subcortical regions negatively correlated with age; brainstem SUVRs correlated only at lower resolution. SUVRs were higher at high resolution across regions. A set of age-dependent, harmonized FDG-PET atlases was developed and the quantitative metrics across age were characterized.