Optimization of the [/sup 18/F]FDG protocol for clinical cerebral studies in 3D PET

The aim of this work was to optimize the [/sup 18/F]FDG cerebral protocol for 3D PET acquisitions on the EXACT HR+. We compared the noise equivalent counts (NEC) and the quantitation of 2D and 3D acquisitions both performed in 7 patients. The 2D and 3D scans were acquired sequentially during respectively 20 and 5 minutes after an injection of 160 to 255 MBq of [/sup 18/F]FDG. A plasma input function was measured in three patients. The NEC analysis indicated that the optimal dose of [/sup 18/F]FDG was close to 200 MBq for a 3D brain acquisition. Using this dose, a 3.5-fold increase in NEC was obtained in 3D compared to 2D allowing thus the acquisition time to be reduced to 6 minutes without any loss of NEC. First results in three subjects indicated slightly higher regional cerebral metabolism rate of the glucose (rCMRglc) in 3D than in 2D: the slope of the linear regression onto the 3D versus 2D rCMRglc values equaled 1.07/spl plusmn/0.008, 1.04/spl plusmn/0.006 and 1.07/spl plusmn/0.005 (p<0.0001). The discrepancy between 2D and 3D values may be due in part to an overestimation of the scatter in 2D and to an undercorrection for the scatter component originating from out of FOV sources in 3D.