Pixel size gradient detector for monolithic crystal PET systems

The use of monolithic crystals to detect the incident gamma photons for PET imaging has the flexibility to virtually define the detector pixel size. These pixels determine the Lines Of Response (LORs) needed during the reconstruction process. Although there are very few groups using monolithic crystals for PET, the most common approach is to homogeneously define the dimensions for the virtual pixels forming a regular grid on the crystal. In this work we present a dimensional pixel gradient method to improve the detector response model. The typical spatial resolution degradation caused by the border effect, is mitigated here using larger pixel sizes towards the crystal edges. This new approach was implemented for the LMEM algorithm using the Tube Of Response (TOR) backprojector. This backprojector identifies the emission probabilities with an approximation of the intersection volume of the TOR within each voxels it passes through. This operator has been modified to calculate the emission probabilities with variable initial and final sizes of each TOR. This method has been tested in two PET prototypes, one of them uses SiPMs and the other uses PSPMTs as photo-detectors. To evaluate the effect of variable virtual pixel sizes in the crystal, the eccentricity and the spatial resolution (according to NEMA standard acquisitions) on reconstructed images of point sources were determined. These results were compared with the LMEM and the MLEM algorithms using homogeneous pixels. The results show similar volumetric resolutions but improvements in the eccentricity. An improvement in more than 50% in eccentricity for the most offcenter radially displaced sources is achieved when the SiPM based system was used. In case of the detector based on PSPMTs, the eccentricity improvement reaches 17%. Finally, this method has been further validated on in-vivo images namely mice injected with NaF using the SiPM-based system and FDG injected patients acquired with a breast dedicated PET based on PSPMTs.