TY - JOUR T1 - <strong>Performance of a Monolithic Crystal PET Detector with Integrated Retroreflectors</strong> JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 3314 LP - 3314 VL - 63 IS - supplement 2 AU - Lemise Saleh AU - Paul Vaska Y1 - 2022/06/01 UR - http://jnm.snmjournals.org/content/63/supplement_2/3314.abstract N2 - 3314 Introduction: The standard commercial positron emission tomography (PET) detector uses pixelated scintillation crystals that provide an average spatial resolution of 4-5 mm. The size of the individual scintillation crystals directly affects resolution, motivating research towards miniature crystals (&lt;=3 mm). This approach can be expensive requiring fabrication of many tiny crystals, and results in substantial dead space. Monolithic scintillators are an alternative though they produce wider light distribution. To minimize light spread, black coatings can be applied to all faces except the photosensor coupled side but this reduces overall light collection efficiency and leads to poorer energy, spatial, and time resolutions. We are implementing an array of trihedral corner reflectors as used in retroreflectors, machined directly on the entrance face of a monolithic scintillator, which can efficiently reflect light back to its origin, and readout by a silicon photomultiplier (SiPM) array. We compare results to previous monolithic approaches including optical coupling of a separate reflective material to the front face of the scintillator, which can result in distortions and reflections due to multiple interfaces and materials with differing indices of refraction.Methods: GATE, a Geant4 toolkit for medical imaging, was used to simulate a 20 mm thick LSO scintillator and a 16 x 16 array of SiPMs coupled with grease (n = 1.5). We compared different entrance face surfaces: 1) fully absorptive (black), 2) flat with a polished finish and coupled specular reflector (Flat Specular), 3) 1 mm retroreflective layer coupled with optical grease (Coupled RR), and 4) 1 mm integrated corner reflector (ICR) array with coupled specular reflector (Figure 1). Surface and reflector finishes were simulated using Davis LUT model. All measurements include realistic photon detection efficiency (40%) and electronic readout noise. To analyze spatial resolution, a perpendicular beam of gamma-rays was moved across the central SiPM in 0.5 mm increments. The true source position was compared to the position centroid and the local slope of these curves were used to convert resolution from bins to mm. To analyze DOI, the light spread function (LSF) at the center of the monolith was characterized using the root mean square (RMS) of the 2D light distribution at the readout plane. Results: For each entrance face finish, the average light distribution of events occurring at the center of the crystal are shown in Figure 2. The black finish only detects photons that are within a detection cone characterized by the critical angle between the LSO (n = 1.8) and optical grease; all other photons reflect off the photosensor and are absorbed by the walls of the crystal. When ICRs are used, significant collection improvements are seen in the central nine photosensors: 94%, 40%, and 26% relative to black, flat specular, and coupled RR, respectively. At the center of the monolith, the ICRs result in an average transverse spatial resolution of 0.85 mm, an improvement of 27% from black, 24% from flat specular, and 15% from the coupled RR finishes (Figure 3 Left). The depth resolutions vary substantially with depth but average to 3.07, 3.26 and 2.78 mm, when using black, flat specular, and coupled RR. In comparison, the ICR detector performs significantly better (average 2.1 mm), and it is more stable across depths (Figure 3 Right). Conclusions: We show that the use of integrated corner reflectors provides a narrower light distribution, while substantially increasing the total number of detected photons, compared to other monolithic approaches. This results in considerable improvements in transverse spatial and depth resolution. Ongoing work includes optimizing the triggering and positioning schemes, as well as determination of the impact of these different designs on energy and time resolution as a function of position within the crystal, in both simulation and experiment. ER -