ABSTRACT
PURPOSE: Reliable 18F-fluorodeoxyglucose (FDG) uptake quantification is crucial for cancer treatment monitoring. While interobserver variability has been found to be lower for a maximum standard uptake value (SUV)max than for an averaged SUV (SUVmean), the repeatability has not been investigated yet. In this study, we determined the repeatability of SUV values in two sequential measurements 5 min apart. METHODS: Positron emission tomography data of malignant chest tumors were acquired dynamically during 45 min in 20 patients. SUV values were derived from the hottest (SUVmax), the mean of the 5 (SUV5) and 10 (SUV10) hottest voxels and the mean of a volume of interest (SUVmean). The repeatability of the SUV measurements was determined as the standard deviation of the difference between the values at 40 and 45 min and represented as Bland-Altman graphs. RESULTS: The standard deviation of the difference between the two sequential scans for SUVmax, SUV5, SUV10 and SUVmean was 1.01, 0.53, 0.37 and 0.28. CONCLUSION: The repeatability of SUV is markedly increased by deriving the value from multiple voxels. Compared to SUVmax, the variability in SUV measurements is reduced by a factor of 2.7 (2.7=1.01/0.37) if 10 voxels are pooled.