Reliability of standardized uptake value normalized to lean body mass using the liver as a reference organ, in contrast-enhanced 18F-FDG PET/CT imaging

Abstract

PurposeTo evaluate the reliability of standardized uptake value (SUV) normalized to lean body mass for maximum (SULmax) and mean values (SULmean) as well as maximum SUV values (SUVmax) in contrast-enhanced 18F-FDG PET/CT by assessment of inter-reader agreement, using the liver as a reference organ. Materials and methods272 images of baseline PET/CT scans were analyzed. A volume of interest (VOI) of 30-mm in diameter was placed by two independent readers in the right liver lobe to measure the parameters. An analysis was performed for the variance, intra-class correlation coefficient (ICC), and Bland-Altman plots. Results and discussionFor Reader 1, the SUVmax ranging from 1.33 to 5.94 (3.20 ± 0.69), SULmean ranging from 0.84 to 3.46 (1.90 ± 0.40) and SULmax ranging from 1.18 to 4.07 (2.38 ± 0.50), were obtained; for Reader 2 the SUVmax ranging from 1.47 to 5.43 (3.20 ± 0.70), SULmean ranging from 0.84 to 3.45 (1.90 ± 0.40), and SULmax ranging from 1.18 to 4.48 (2.38 ± 0.50), were obtained. The coefficient of variance for SUVmax, SULmean, and SULmax, were 21.9%, 21.1%, and 20.8%, respectively, having no significant differences between Reader 1 and Reader 2. The ICC of the two readers for SUVmax, SULmean and SULmax were 0.975, 0.982 and 0.977 (95% CI of 0.97, 0.98 and 0.97; p < .001) respectively. Bland-Altman plots revealed that SUVmax gave the best agreement with 97.1% of measurements falling within 2SD. ConclusionThere is an excellent inter-reader agreement for liver SUVmax, SULmean and SULmax, and the best reliability of measurements achieved with SULmax in contrast-enhanced PET/CT scans.