Abstract
387
Introduction: We’ve developed a segmented light guide, Prism-PET, specialized for multicrystal single-ended readout depth-encoding PET modules. Our light guide enhances depth-of-interaction (DOI) resolution and crystal identification accuracy, enabling up to 9-to-1 crystal-to-readout coupling. In this work, we demonstrate the capabilities of Prism-PET using Monte Carlo simulations and experimentally acquired flood data.
Methods: We simulated PET data using the Monte Carlo optical ray tracing software TracePro. Three detector modules, each consisting of a multicrystal scintillator array with a light guide coupled at the entrance layer and an 8 x 8 array of 3.2 x 3.2 mm2 silicon photomultiplier (SiPM) readout pixels coupled to the opposite end, were simulated: (1) a 4-to-1 coupled module (16 x 16 array of 1.4 x 1.4 x 20 mm3 LYSO crystals with 0.2 mm gaps) with a 1 mm thick uniform glass light guide, (2) a 4-to-1 coupled module (same scintillator array as in case (1)) with a Prism-PET light guide, and (3) a 9-to-1 coupled module (24 x 24 array of 0.96 x 0.96 x 20 mm3 LYSO crystals with 0.1 mm gaps) with a Prism-PET light guide. 40 uniformly distributed 511 keV gamma ray photoelectric absorptions were simulated per crystal, resulting in 10,240 events in both 4-to-1 modules and 23,040 events in the 9-to-1 module. We fabricated and acquired flood data on three detector modules equivalent to the ones simulated in TracePro. Barium sulfate (BaSO4) was used to fill the intercrystal spaces and act as a diffuse reflector in the crystal arrays and light guides. All crystals were fully polished and the modules were wrapped in black tape. A 3 MBq Na-22 source was used to acquire flood data. Photopeak analysis was done for each crystal via visual inspection of the energy distributions. 10,000,000 events in each of the 4-to-1 modules and 22,500,000 events in the 9-to-1 module were used for analysis. We also acquired flood data at 5 different depths (2, 6, 10, 14 and 18 mm) in each module using lead collimation with a 1 mm pinhole to determine the effect of light guide on DOI resolution. Energy resolution was calculated for both the uncollimated and lead collimated datasets. Energy weighted average method was used to perform crystal identification and DOI localization of each simulated and experimentally acquired gamma ray event.
Results: Simulated and experimental flood data showed strong agreement for all three detector modules. Flood histograms for the glass light guide module have good crystal separation at the center and poor separation at the edges and corners. Both Prism-PET light guides have excellent crystal separation throughout the crystal arrays, including at the edges and corners. Average energy resolutions across all crystals for the 4-to-1 glass, 4-to-1 Prism-PET and 9-to-1 PrismPET modules were 20%, 14% and 16% without DOI collimation and 13%, 9% and 10% with DOI collimation, respectively. DOI resolutions based on experimental results were 5 mm FWHM and 2.5 mm FWHM for the 4-to-1 coupled glass light guide and Prism-PET modules, respectively.
Conclusions: We’ve shown the performance capabilities of single-ended readout depth-encoding detector modules with Prism-PET. Cost-effective high resolution PET systems, including human organ-dedicated and total-body scanners, can be developed using Prism-PET. Acknowledgments: We would like to thank the NIH for funding this project (R21 EB024849) and our colleagues at PETsys Electronics SA. Figure 1. (a) Schematic diagram of a 4-to-1 coupled Prism-PET detector module. (b) Fabricated 4-to-1 coupled Prism-PET module. (c) TracePro simulation of a single photoelectric event in the 3 simulated modules. (d) Experimental flood histogram results showing Prism-PET’s enhanced crystal identification performance. (e) Crystal separation in the x-direction for a center, edge and corner pixel. (f) Energy resolution with (black) and without (colored) DOI collimation. (g) DOI resolutions of the 4-to-1 coupled uniform glass and Prism-PET modules.