Impact of axial collimation on hybrid PET image quality

Previous studies have suggested that novel axial collimator designs may be able to increase coincidence efficiency, and better shield out-of-FOV singles photons. It was found that increasing the air gap between the septa and using converging septa increased the sensitivity to the central FOV and improved the shielding from out-of-FOV events. We are currently investigating image quality of such collimator designs in relation to rebinning, and reconstruction algorithms. The GEANT Monte Carlo simulation code was used to simulate high count acquisitions with varying collimator designs: axial parallel or converging septal collimators and different septal spacing. The data were reconstructed using 2D single slice rebinning (SSRB) followed by filtered back projection (FBP). Artifacts resulting from the shadows cast by septa were visible when using 2-cm and 4-cm septal spacing, but not 1-cm septal spacing. It was observed that simulating a typical noise level into the projection data caused the shadow artifacts to be visually suppressed for 1-cm and 2-cm septal spacing, but not for 4-cm spacing. In addition, it was observed that 3D Butterworth post-filtering also suppressed shadow artifacts, except for the 4-cm septal spacing case.