Comparison of PET/CT SUV metrics across different clinical software platforms

Abstract

ObjectiveTo assess the agreement of SUV metrics across the different clinical PET reading software platforms available at our institution. MethodsPET/CT images were reviewed on four different FDA-approved software platforms: syngoMMWP VE36A and syngo.via VB30A (Siemens), Intellispace Portal 9.0 (Philips), and Encore 6.7 (MIM Software). A total of thirty SUV measurements were derived from ten 18F-FDG PET/CT oncology studies. A volume of interest (VOI) was drawn around the primary tumor to determine lesion SUVmax and a 3 cm diameter spherical VOI was placed in the right lobe of the liver to determine liver SUVmean and liver SUVmax. ResultsFor lesion SUVmax, statistically significant differences were found for syngoMMWP VE36A vs syngo.via VB30A (p = 0.002), syngoMMWP VE36A vs Intellispace Portal 9.0 (p = 0.002), and syngoMMWP VE36A vs Encore 6.7 (p = 0.001), respectively. For liver SUVmax, a statistically significant difference was found for syngoMMWP VE36A vs syngo.via VB30A (p = 0.033) only, whereas for liver SUVmean, no statistically significant differences were determined. A small systematic bias was found between syngoMMWP VE36A and all other platforms for lesion SUVmax. ConclusionSignificant differences and systematic biases were observed when measuring lesion SUVmax using different reader software systems. Although these differences may not be clinically significant, this bias could confound outcomes for quantitative, precision-research protocols. Hence, it is important for nuclear medicine departments to take SUV metric agreement into consideration, especially when transitioning to a new clinical platform.