A Comparison of Five Whole-Body PET Scanners by Scanning Hoffman Brain Phantom

We compared the performance of three PET/CT scanners from GE, the CTI ECAT EXACT HR+, and the prototype MDAPET camera using the Hoffman brain phantom. GE cameras were the new BGO based Discovery STE PET/CT (DSTE), the LYSO based Discovery RX PET/CT (DRX), and the BGO based Discovery ST PET/CT (DST). All of these cameras are based on block design where each block is an array of many crystals. Each crystal is 4.7 times 6.3 times 30 mm for DSTE, 4.2 times 6.2 times 30 mm for DRX, 6.3 times 6.3 times 30 mm for DST, 4.39 times 4.05 times 30 mm for EXACT HR+, and 2.7 times 2.8 times 15 mm for MDAPET in transaxial, axial, and radial directions, respectively. MDAPET has smaller axial FOV of 3.8 cm compared to other four cameras with axial FOV of 15.5 to 15.7 cm. All data were taken in 3D mode. For three GE cameras CT data were used for attenuation correction, for HR+ transmission data were used, and for MDAPET attenuation corrections were calculated. For each camera, data were processed using manufacturer-supplied software and the default image reconstruction parameters. Images were evaluated by visual comparison of PET images with corresponding CT images and phantom true pattern by human observers, and also by comparing profiles running through similar image slices. All cameras reproduced the main structures of the Hoffman brain phantom; however, there were differences in reproduction of the fine structures of the phantom and there were also differences in profiles. DRX and DSTE cameras performed slightly better in comparison to DST and there were no significant differences between their results and HR+ in terms of spatial resolution. Overall, the fine structures of the brain phantom were best observed in images acquired with MDAPET camera which has smaller BGO crystals.