PURPOSE: Recent developments in the design of positron emission tomography (PET) scanners have made three-dimensional (3D) data acquisition attractive because of significantly higher sensitivity compared to the conventional 2D mode (with lead/tungsten septa extended). However, the increased count rate in 3D mode comes at the cost of increased scatter, randoms, and dead time. Several schemes to correct for these effects have been proposed and validated in phantom studies. In this study, we evaluated the overall improvement afforded by 3D imaging in quantitative human brain PET studies carried out at our institution.METHODS: Subjects were studied using sequential/interleaved 2D and 3D data acquisition with a GE Advance scanner. We calculated regional and global cerebral glucose metabolism with [(18)F]flourodeoxyglucose (FDG) and estimated rate constants for striatal [(18)F]fluorodopa (FDOPA) uptake.RESULTS: FDG: Global mean glucose metabolic rates were in almost complete agreement (within 1%) between the two modes whereas the regional differences ranged from -7.7% to +9% for all cortical structures. However, for small regions (<2 cm(2)) like caudate nuclei, the maximum difference was 14.7%. FDOPA: A significant improvement in image quality was evident in 3D mode and there was complete agreement between the estimated parameters in the two scanning modes for the same noise equivalent counts: Striatal-to-occipital ratio (SOR) and striatal FDOPA uptake (K(i)(FD)) had mean differences of less than 2% and 5%, respectively.CONCLUSIONS: 3D FDG studies can be done with either half the injected dose or half the scan duration to a comparable 2D study. 3D PET imaging has distinct advantages over 2D in the quantitative fluorodopa studies.