A comparison of three PET phantoms for evaluating spatial resolution

2512 Objectives PET spatial resolution impacts quantitation, which is important in multicenter trials. Phantoms are used to evaluate spatial resolution for both clinical acceptance testing and scanner qualification in clinical trials. The purpose of this work was to compare imaging results of three phantoms on different PET/CT systems. Methods The SNMMI CTN phantom, ACR phantom, and EANM NEMA NU-2 phantoms were imaged multiple times on a GE Discovery ST, GE Discovery 710 ,and Siemens Biograph mCT. Each acquisition was 4 minutes per bed position. Images were reconstructed using harmonized parameters to produce comparable quantitative results. Advanced reconstruction methods such as time of flight and point spread function recovery were not used. The results of the multiple scans were averaged. SUVmean was measured using 2D ROIs sized to fit the known diameter of each lesion. Finally, recovery coefficient curves were generated plotting the measured SUV divided by the true SUV of each lesion and compared to show all three phantoms on each PET/CT system. Results Selected reconstruction harmonization parameters resulted in similar resolution recovery curves across scanners. The NEMA NU-2 and CTN phantoms both have spherical lesions and recovery curves were most similar with these two phantoms. Relatively higher recovery values were observed from the ACR phantom, since this employs rod shaped lesion compartments. Compared to phantoms with spherical objects, the ACR phantom may overestimate recovery values of spherical objects. Conclusions All 3 phantoms demonstrate similar trends in underestimating SUVs for smaller object sizes due to the partial volume effect. While the NEMA NU-2 and CTN phantoms result in comparable values, it is important to recognize the ACR phantom will result in relatively higher recovery curves due to the geometry of the rods versus spherical inserts.