Design study of future 3-D PET systems

Current detector technology based on BGO block-detectors is capable of achieving an intrinsic resolution of 3-4 mm. In order to fully utilize this high resolution, 3-D acquisition techniques are necessary resulting in an improved sensitivity of 6-7, compared to 2D technology. However, dead time problems start to get severe at administered doses 3-5 times lower than previously met in 2-D. In this work, the trade-off between the sensitivity increase given in 3-D at different system geometries and the noise increase due to system count rate limitations and random coincidences is studied, Based on Monte Carlo simulations, the amount of true, scatter and random coincidences have been estimated within a wide range of geometries. With a simple dead time model for the data collection system we have investigated if an optimized geometry can be found for brain function studies. The optimization is, as expected, not unique but is strongly dependent on the administered amount of activity as well as the choice of the dead time parameters