Reliability of SUVmax vs. SUVmean in FDG PET/CT

2121 Objectives To compare the accuracy and reliability of SUV(b.w.)max vs. SUV(b.w.)mean in oncologic FDG PET pts. Methods 278 FDG-positive tumoral lesions detected in 150 consecutive pts who had undergone a PET/CT were considered. The PET/CT was acquired 1 hour after the injection of 2.2 MBq/kg/b.w. of FDG by a GE Discovery STE hybrid tomograph. Lesion size ranged 2 to 6 cm, mean +/- 1 SD = 2.8 +/- 1.3. SUV(b.w.)max and SUV(b.w.)mean were measured in each lesion (10 measurements per lesion) obtaining a total of 2,780 measurements. For the purpose of SUV measurement an automatically generated planar 96 mm2 ROI was systematically used. CT (140 kV, mA 50) was used for attenuation correction. Images were reconstructed by OSEM method. The variations in SUV(b.w.)max and SUV(b.w)mean measurements for each lesion (1 SD and range) were expressed as the percentage variations in comparison to the mean value. Results In our series the variability in SUV(b.w.)max measurement was +/- 2% (1 SD) with a range from -7% to + 7%, while the variability in SUV(b.w.)mean measurement was +/- 7% (1 SD) with a range from – 23% to + 23% (p Conclusions Accordingly with our data, the SUVmax measurement appears to be a more reliable and accurate parameter than SUVmean for the semi-quantification of the FDG PET/CT. Therefore, SUVmax should be preferred especially when comparing serial FDG PET/CT examinations performed in the same patient to evaluate the response to treatment.