A phantom study evaluating the quantitative aspect of 3D PET imaging of the brain

Phantom studies are used to develop a reliable quantitative data processing protocol for 3D PET brain scanning for conditions typically encountered in FDG and neuroreceptor brain imaging. These protocols often span several half-lives of the injected radiotracer thus resulting in a greatly varying statistical content of the acquired data over the study duration. Detector normalization, scatter correction and their interplay over a wide range of statistical content of acquired data were evaluated. Overall sensitivity calibration factors were determined after all other quantification corrections were applied to the data. The result is an optimum data processing protocol that includes an iterative convolution subtraction scatter correction method, a normalization procedure that takes into account the geometric properties of the scanner and a region of interest based calibration procedure, applied in this order. This protocol yields a 3D PET quantification accuracy within approximately 3% of independently measured concentration values for scanning conditions that include variation in the number of acquired counts from one million to several hundred millions and variation in size and shape from a 20 cm diameter phantom to a tapered phantom with minimum cross section of 3.7 x 14.5 cm2. This performance is comparable with that of the 2D acquisition mode.