Simulation of countrate performance for a PET scanner with different degrees of partial collimation

We present a simulation study of the global countrate performance of a PET scanner with different degrees of partial collimation. Partial collimation is achieved by symmetrical removal of 10 to 90% of septa from the standard 2D septa set for the GE Advance PET scanner. Results are evaluated in terms global count rates and noise equivalent counts (NEC) as well as the scatter fraction. We study system behavior with a photon tracking simulation package (SimSET). It is calibrated to measured data for 2D and fully-3D acquisition modes using the NEMA NU-2 countrate phantom. We find good agreement between the measured and simulated count rates for the Advance PET scanner in 2D (full collimation) and fully-3D (no collimation) acquisition modes. Partial collimation, particularly with one half or two-thirds of septa removed, shows significant improvement in statistical data quality over 2D or 3D acquisitions for whole-body imaging in clinically important activity regions, especially for larger phantoms. In order to turn this statistical advantage into improved PET image quality, correction and reconstruction schemes suitable for the partial collimation data have to be developed