TY - JOUR T1 - <strong>From a PMT based to a SiPM based PET system: NEMA NU 2-2018 performance of the digital Biograph Vision 450 and a study to define matched acquisition/reconstruction parameters</strong> JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 141 LP - 141 VL - 61 IS - supplement 1 AU - Thomas Carlier AU - Yanic Bercier AU - Caroline Bodet-Milin AU - Caroline ROUSSEAU AU - Maurizio Conti AU - Bernard Bendriem AU - Francoise Kraeber-Bodere AU - Ludovic Ferrer Y1 - 2020/05/01 UR - http://jnm.snmjournals.org/content/61/supplement_1/141.abstract N2 - 141Objectives: The use of silicon photomultipliers (SiPMs) offers many advantages as a photodetector including compactness, improved time-of-flight resolution and finer spatial decoding scheme enabled by smaller crystal size. However, few questions arise when moving from PMT to SiPM based PET system as far as acquisition parameters is concerned. The purpose of this study was firstly to derive the NEMA NU 2-2018 performance (NEMA) of the newly released Biograph Vision 450 PET/CT and, secondly, to propose an approach based on noise measurement to adapt present clinical acquisition and reconstruction parameters to the new digital PET systems. Methods: All measurements were conducted on the Biograph Vision 450 (Vision) and Biograph mCT with TrueV (mCT). Spatial resolution, sensitivity, scatter fraction, noise equivalent count rate (NECR), time-of-flight resolution, image quality and co-registration accuracy were assessed following the NEMA procedure for Vision only. The adaptation of acquisition and reconstruction parameters from mCT to Vision was done using NEMA image quality phantom (although using a non-NEMA processing approach described below) with a sphere-to-background contrast of 4:1. The phantom was first imaged on Vision for 240 s and then immediately imaged on mCT for 250 s to account for the decay. For this purpose, the noise level (defined as image roughness) reached using mCT was set as the reference value for 6 different number of net true coincidences (time acquisition: from 4 to 0.5 min). The noise level computed using Vision was matched to the reference noise level (within 0.01%) using different reconstruction set-up (matrix size and reconstruction parameters) to determine the potential reduction of counts number for the same noise level. The 3D contrast recovery (CR, obtained from a 3D VOI) for the hot and cold spheres was also derived for each final set-up so that a comparison in terms of contrast could be also achieved. The mCT reference reconstruction parameters for the 3D OP-OSEM+TOF+PSF used in routine clinical practice in our institution are 3 iterations and 21 subsets using a matrix size of 200×200. A post-filtering was not applied to ease the results interpretation. Results: Vision sensitivity was 9.1 kcps/MBq at the center of the field-of-view for a timing resolution of 213 ps at 5.3 kBq/mL. At peak NECR, scatter fraction was 37.5% and the count rate was 160 kcps. The co-registration accuracy was 1.12 mm and the NEMA-based CR for the 10mm sphere was better than 75% regardless the reconstruction set-up studied. The mCT reference noise properties could be achieved using Vision with a scan time reduction (STR) of 1.34 with 4 iterations and a 440×440 matrix size (or STR=1.89 with a 220×220 matrix size) together with a 3D CR improvement of 53% for the 10mm sphere. Conclusions: The Vision exhibited improved NEMA performances compared with mCT. Using the proposed approach, the time acquisition could be divided by almost 2, while keeping the same noise properties as that of the mCT with a marked improvement of contrast recovery. ER -