Design optimization using GATE Monte Carlo simulations for a sub-0.5 mm resolution PET scanner with 3-layer DOI detectors

<i>Objective.</i>For rodent brain PET imaging, spatial resolution is the most important factor for identifying small brain structures. Previously, we developed a submillimeter resolution PET scanner with 1 mm crystal pitch using 3-layer depth-of-interaction (DOI) detectors. However, the spatial resolution was over 0.5 mm due to a relatively large crystal pitch and an unoptimized crystal layer design. Here we use Geant4 Application Tomographic Emission (GATE) Monte Carlo simulations to design and optimize a sub-0.5 mm resolution PET scanner with 3-layer DOI detectors.<i>Methods.</i>The proposed PET scanner has 2 rings, each of which has 16 DOI detectors, resulting in a 23.4 mm axial coverage. Each DOI detector has 3-layer lutetium yttrium orthosilicate crystal arrays with a 0.8 mm crystal pitch. We employed GATE Monte Carlo simulations to optimize three crystal layer designs, A (4 + 4 + 7 mm), B (3 + 4 + 4 mm), and C (3 + 3 + 5 mm). Spatial resolution and imaging performance were evaluated with a point source and resolution phantom using analytical and iterative algorithms.<i>Main results.</i>Among the three designs, design C provided the most uniform spatial resolution up to the radial offset of 15 mm. The 0.45 mm diameter rod structures were resolved clearly with design C using the iterative algorithm. The GATE simulation results agreed with the experimental data in terms of radial resolution except at the radial offset of 15 mm.<i>Significance.</i>We optimized the crystal layer design of the mouse brain PET scanner with GATE simulations, thereby achieving sub-0.5 mm resolution in the resolution phantom study.