Quad-HIDAC PET: comparison of four image reconstruction techniques for high resolution imaging

The quad-HIDAC PET scanner offers unparalleled high resolution PET imaging capability (<1 mm), combined with notable count rate performance (400 kcps) and sensitivity (up to 2% with scatter correction). To some extent the image resolution realized by the system is determined by the choice of image reconstruction algorithm. This work compares 4 reconstruction techniques applied to data from the quad-HIDAC scanner: 2 iterative methods (ROPLE and 3D OSEM) and 2 analytic methods (BPF and 3D RP). These four methods were chosen so as to include both a list-mode based and a projection-based method in each category. The algorithms were compared in terms of contrast, noise and resolution for a specially designed cylindrical phantom. The results indicate that the iterative methods offer improved resolution and contrast for a given noise level compared to the analytic methods. Of these iterative methods, the new list-mode technique (ROPLE) performed better than the projection-based technique.