PT - JOURNAL ARTICLE AU - Diniz Sá AU - Nuno Ferreira AU - Pedro Correia AU - Pedro Encarnação AU - Joana Menoita AU - Ana Silva AU - Fabiana Ribeiro AU - Filipe Castro AU - Fabiana Rodrigues AU - Cristina Santos AU - Carlos Ramos AU - Joao Veloso TI - 3D List-mode Image Reconstruction for the EasyPET prototype DP - 2018 May 01 TA - Journal of Nuclear Medicine PG - 1785--1785 VI - 59 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/59/supplement_1/1785.short 4100 - http://jnm.snmjournals.org/content/59/supplement_1/1785.full SO - J Nucl Med2018 May 01; 59 AB - 1785Objectives: To present the first results of 3D iterative image reconstruction algorithms developed for EasyPET, a 3D PET prototype with variable spatial sampling. The easyPET is a low-cost rotating PET scanner aimed primarily at education and pre-clinical imaging. A 2D version was previously developed [1]. In this work, we used a 3D version with 2 detector heads, each consisting of a single row of 16 crystals along the axial direction (see supporting material, fig. 1). With rotation, data from 16 direct planes can be simultaneously acquired with 2 opposing crystals per plane, forming a total of 16x16=256 oblique planes between crystals. Rotation using two motors allows to have a cylindrical Field Of View (FOV) of variable size, up to 50 mm in diameter (35 mm axially), that can be sampled with different degrees of detail, depending on the size and number of the rotation steps used. Fans with up to 3290 Lines-Of-Response (LORs) in steps of 0.031° can be measured for each one of the 800 positions (steps of 0,45°) in a full circle (fig 2). The sensitivity of about 0.05% is reduced due to the limited number of crystals (the exact value depends on the size of transaxial FOV). Methods: We implemented a 3D list mode iterative MLEM and OSEM algorithm and assessed the results relative to the STIR reconstruction library [2] using different test objects: individual and groups of point sources, uniform cylinder, cylinder with 2 cavities and other phantoms. We also acquired images of a mouse injected with 18F-FDG. The reconstruction software can use different criteria to calculate the contribution of each voxel to a LOR: based on the distance of a point to the line of Response (LOR) joining the two detectors in coincidence and based on the length of the LOR in the voxel. Data subsets are also implemented. Results: A spatial resolution of 1.1 mm, uniform in the transaxial FOV, was measured following the NEMA NU 4-2008 standard. The supporting material shows examples of the system (Figs. 1-2) and images obtained with the implemented algorithms (Fig. 3). Comparison with images reconstructed with STIR helped in the validation of the algorithms, which are comparable in detail and noise. The algorithms’ performance with different spatial sampling parameters is currently being evaluated to define the best set of parameters for imaging mice. Conclusions: The easyPET system is a low-cost approach to the acquisition of high spatial resolution 3D PET images. Although its sensitivity is much smaller than that of a full-ring 3D scanner, its flexible spatial sampling can in principle provide images with very good quality and spatial resolution in environments where acquisition time or injected activity is not critical. References:[1] V. Arosio, M. Caccia, I.F. Castro et al., “easyPET: a novel concept for an affordable tomographic system”, Nuclear Instruments and Methods in Physics Research A, 845, pp. 644-647, 2017 [2] Kris Thielemans, Charalampos Tsoumpas, Sanida Mustafovic et al. “STIR: Software for Tomographic Image Reconstruction Release 2”, Physics in Medicine and Biology, 57 (4), 2012 pp.867-883.