Digital PET systems performances: A multicenter phantom based comparison in clinical environment

193 Objectives: Recently, a third manufacturer proposed a positron emission tomography system based on silicon photomultiplier, commonly called “digital pet”. This technology provides a better sensitivity and a lower and more robust time resolution, leading to a better signal to noise ratio. A better spatial resolution thanks to smaller crystal units or more precise optic detectors was also demonstrated. The purpose of this work was to compare those state of the art digital pet systems thanks to phantom examinations using clinical-like acquisition and reconstruction protocols. Methods: We compared five systems installed in 5 different French nuclear medicine departments. Philips Vereos (TEP1), GE DMI 4rings (TEP2), Siemens Vision 600 (TEP3 and TEP4). We acquired a NEMA IEC phantom (0.7 to 26.5 mL hot spheres) and a Jaszczak phantom equipped with at least 5 sub-centimeter spheres (31, 63, 125, 250 and 500 µL) and cold rods (rod diameters were 4.8, 6.4, 7.9, 9.5, 11.1 and 12.7 mm). For the IEC phantom, the activity at acquisition time was 50 MBq with a 4:1 contrast, and for the Jaszczak phantom, 20 MBq with 3:1 and 6:1 contrast. IEC phantom was acquired with one bed position during 600s. Jaszczak phantom was acquired in list mode during 600s and reconstructed in 10, 5, 3, 2 and 1 min emission time. We considered reconstruction parameters currently used in each center for whole body clinical examinations. IEC phantom was used to measure sensitivity, signal to noise ratio in homogeneous area and contrast recovery from max and mean value. Jaszczak phantom was used to evaluate the visual detectability: for each emission time, 2 readers assigned a score (0= invisible, 1= slightly visible, 2= visible) to the 4 smallest cold rods areas, and to the four smallest visible hot spheres (63, 125, 250 and 500 µL for 3:1 contrast and 31, 63, 125 and 250 µL for 6:1 contrast). The scores were averaged to obtain a detectability score as a function of emission time (max=16). Results: Considering IEC phantom, sensitivity was 3.1, 8.52, 9.38 and 9.66 cps/MBq, Signal to noise ratio was 16.6, 36.3, 35.8 and 28 for TEP1 to 4 respectively. Contrast recovery from max value were [0.56;0.79;0.98;1.07;1.13;1.22], [0.59;0.88;1.08;1.07;1.12;1.1], [0.78;1.09;1.2;1.2;1.21;1.21], [0.81;1.18;1.15;1.2;1.22;1.21] for TEP1 to 4 respectively. Contrast recovery from mean value were [0.42;0.56;0.7;0.73;0.84;0.96], [0.42;0.57;0.69;0.75;0.82;0.84], [0.52;0.64;0.77;0.83;0.89;0.94], [0.51;0.66;0.77;0.85;0.92;0.95] for TEP1 to 4 respectively. Considering the detectability score on Jaszczak phantom, we found with 6:1 contrast from 1 to 10 min emission time [3;4.5;6;7.5;8.5] [4;7;8.5;10.5;11], [9;11;11.5;13;14.5] and [10;12.5;14.5;16;16] for TEP1 to 4 respectively. With 3:1 contrast we found [2;3;3.5;6;7.5], [4.5;6.5;9;9.5;11] and [6;9;10;12;13.5] for TEP1 to 4 respectively. Conclusions: SiPM technology leads a substantial improvement in the quantification and detectability of small lesions. In this work we focused on clinical locally used reconstructions, and we found variability between similar systems due to variation of reconstruction parameters. In a future work, we plan to measure those indices on a larger amount of devices, including non-SiPM systems, to have a more accurate view of their actual performances.