3D Impact Position Determination Using Neural Networks in the IMAS Total-Body PET

Total-Body Positron Emission Tomography (TB-PET) technology has seen a significant rise in popularity in recent years, due to their increased sensitivity with respect to Whole-Body (WB) PET systems. This is mainly attributed to their extended axial Field of View (FOV) and, in a few cases, the capability of Time of Flight (TOF) information. This combination enables simultaneous assessment of the kinetics processes across multiple organs. We have recently developed a TB-PET scanner, named IMAS, based on semi-monolithic LYSO crystals. This geometry combines the good 3D spatial resolution of monolithic-based detectors, including Depth of Interaction (DOI), while maintaining the good energy and timing performance of pixelated scintillators. To accurately determine the interaction position in the monolithic and DOI directions, a positioning method based on Neural Networks (NNs) was employed. The NNs were trained using experimental data acquired from a small reference set of detectors, since collecting training data from the full scanner would have last very long. To extend this positioning approach to the entire system, a slab categorization procedure was applied to match the behavior of all slabs to the reference set. The results obtained in this study show a system spatial resolution of 4.4 mm, 4.3 mm and 2.9 mm at 30 cm from the center of FOV for the radial, tangential and axial directions, respectively.