Abstract
1502
Introduction: We’ve developed PET detector modules with single-ended depth-encoding readout using prism-mirror light guide array as efficient 180o light bending reflectors. Our novel Prism-PET modules provide a practical and cost-effective approach for the construction of high resolution and high sensitivity PET scanners with good depth-of-interaction (DOI) corrected energy and timing resolution. In this work, we show simulation results of the performance of Prism-PET brain scanner with depth-encoding.
Methods: Figure (a) shows the process flow for image reconstruction using depth-encoding readout. We utilized the open-source GEANT4 application for tomographic emission (GATE) to simulate the performance of Prism-PET brain scanners and quantify their sensitivity and spatial resolution. Figure (b) shows GATE model of the Biograph Vision (left) and our proposed Prism-PET (right) brain scanner, approximately scaled to their relative size. Insets in Fig. (b) show the 3D view of one Prism-PET module (16x16 array of 1.5x1.5x20 mm3 crystals coupled to prism-mirror light guide array) and schematic representation of depth-encoding method (by redirecting photons that travel towards the prisms to neighboring crystals which are coupled to neighboring readout pixels). Both Biograph Vision and Prism-PET scanners used block-SiPM geometries and all simulations used fluorine-18 (18F) positron source where the radioactive decay, the positron range, and the acollinearity were modeled. Figure (c) summarizes the design characteristics of all three PET scanners considered in this work. Absolute and axial sensitivities were calculated based on the NEMA-NU4 procedure, with a 18F spherical source of 0.3 mm diameter centered in the FOV and then moved to several points in the axial direction. An activity level of 100 kBq was used with 5,000,000 particle events simulated. Activity levels, particle counts and particle cuts, were varied to ascertain convergence in the results. Both the Biograph vision and Prism-PET modules had an energy window of 450-650 keV, with a 320 ns paralyzable dead time and 4.5 ns coincidence window used. Energy resolution of 15% at 511 keV was modeled, along with LYSO noise (caused by the intrinsic radiation of Lutetium). For evaluating the imaging performance, a 4-size hot spot phantom with rod diameters of 3.2 mm, 2.4 mm, 1.6 mm and 1.0 mm was used. Each had an 18F positron source with a uniform activity concentration of 20 kBq/cc. An acquisition time of 480 seconds was used for these simulations. Reconstruction was done with CASToR using the ROOT output from gate. The incremental Siddon projector was used with 30 MLEM iterations and 1.0 x 1.0 x 1.0 mm3 voxels. Images were then viewed in AMIDE with trilinear interpolation and MIP rendering, with no other post processing steps.
Results: Figure (d) shows the axial sensitivity comparison where the highest absolute sensitivity of 10.7% is obtained for Prism-PET with 4-to-1 scintillator-to-SiPM coupling. Overall, Prism-PET scanners can achieve very high sensitivity due to smaller ring diameter. The sensitivity of Prism-PET can be further improved by increasing both the axial coverage and the geometric packing (increase number of blocks per ring to reduce gaps). Figure (e) shows the hot spot resolution phantoms where Prism-PET with 9-to-1 scintillator-to-SiPM coupling can resolve 1.0 mm spots. In addition, the high spatial resolution of the Prism-PET scanner is maintained across the entire FOV due to depth-encoding (Fig. (f)) and reduction of the parallax effect.
Conclusions: PET scanners based on our detector modules with prism mirror light guide arrays can achieve high and uniform spatial resolution (9-to-1 coupling of 1 mm crystals), high sensitivity (20-mm thick detectors, and inter-crystal Compton scatter recovery), good energy and timing resolution (using DOI-correction), and compact size (DOI encoding eliminates parallax error and permits smaller ring-diameter).
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.