Room: 304ABC
Purpose: This work aims to investigate an all-side readout scheme for a continuous scintillator based detector for positron emission tomography (PET), which not only improves the detector positioning performance, but also makes the continuous crystal based detector more practical to use for a PET system.
Methods: The proposed readout scheme uses 2×2 SiPM arrays (16×16 mm² size of each SiPM) on 6 polished surfaces of a scintillator cubic (32×32×32 mm³), with total 24 readout channels. A conventional readout scheme, using an 8x8 sensor array (4×4 mm² size of each SiPM) coupling to one surface (leaving 5 other surfaces coated in black), was used as a comparison. GATE simulations were conducted to study the scintillation light propagation and collection for 511-keV gamma interactions without the effect of Compton scattering. For both cases, artificial neural networks with identical hidden layer structure were used as positioning estimators. The two readout schemes were evaluated and studied by sample events evenly distributed on the crystal’s central plane.
Results: The average position precision (in FWHM) of the proposed scheme were 0.57 mm while for the conventional scheme was 1.13 mm. The average shift between the calculated and actual interaction positions at the crystal edge areas (distance from the edge <4 mm) was reduced from 0.8 mm to 0.1 mm.
Conclusion: The simulation of proposed readout scheme showed good and uniform performance over the whole test area. Considerable improvement in the detector’s positioning performance in edge areas can be obtained over the conventional design. With symmetrical geometry design, same positioning performance can be expected on z-direction. Moreover, the scheme doesn’t require complex surface treatment, which not only makes the assembly easier for implementation but also simplifies the optical propagation model. In summary, the proposed readout scheme showed very good potential for developing a continuous scintillator-based PET detector.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by 1R01EB019438, 1R01CA218402, R21CA187717.