Impact of Inter-software and Inter-reader Agreement on FDG PET/CT Liver SULmean

Objectives To evaluate inter-software and inter-reader agreement and how the site of the volume of interest (VOI) affect the agreement or variability of liver mean standardized uptake value normalized to lean body mass (SULmean) on FDG PET/CT scans

Methods PET/CT studies from 200 patients (107 males and 93 females) with mean age 51.1 ± 21.0 yrs who had normal livers by structural imaging were included in the study. The average uptake time was 65.7 ± 8.0 min. SULmean was measured with a 30-mm- diameter spherical VOI placed within the right lobe of the liver (above, below, and at the level of the main portal vein) by two readers independently, using three workstations: GE AW, MimVista, and Mirada, for a total of 18 measurements for each PET/CT scan. Analysis of variance, intraclass correlation coefficient (ICC) analysis, and Bland-Altman analysis were performed

Results SULmean was between 1.44 and 1.55 at the upper, portal, and lower levels of the right lobe of the liver measured at the three workstations. The overall ICC for single readings on all 18 measurements of liver SULmean was 0.929. The ICC of the two readers varied between 0.917 and 0.977 at each level for the three softwares. The ICC between the softwares varied between 0.920 and 0.970 at each level for the two readers. The coefficient of variation (COV) of liver SULmean as measured by the different workstations varied between 3.26% and 5.43% at each level for the two readers. Mean absolute bias between the softwares was at most 0.04 with 1.96 SD absolute value not exceeding 0.21

Conclusions Liver SULmean at FDG PET/CT has excellent inter-reader agreement, with similar values and variance whether measured at the upper, lower, or portal vein levels. The inter-software agreement is very good with absolute differences in liver SULmean not exceeding 0.21 and COV < 5.43%. This is within variations acceptable by PERCIST 1.0 criteria