Harmonization of Standardized Uptake Value among Different Generation PET/ CT Cameras Based on a Phantom Experiment —Utility of SUVpeak—

Abstract

Standardized uptake value (SUV) has been widely used as a semi-quantitative metric of uptake in FDGPET/ CT for diagnosis of malignant tumors and evaluation of tumor therapies. However, the SUV depends on various factors including PET/CT scanner specifications and reconstruction parameters. The purpose of this study is to harmonize the SUV among two PET/CT models of different generation: two units of Discovery ST Elite Performance(DSTEP) and Discovery 690 (D690) PET/CT scanners. The NEMA body phantom filled with 18F solution was scanned for 30 minutes in list-mode. The D690 PET images were reconstructed with OSEM, OSEM+TOF, and OSEM+PSF. Gaussian post-filters of 4-9 mm FWHM were applied to find the parameters that provides harmonized SUV. We determined the SUV-harmonized parameter for each reconstruction algorithm. Then, the 10 PET images simulating clinical scan conditions were respectively generated to evaluate the bias and variability of SUVmax and SUVpeak. The SUVmax strongly depended not only on spatial resolution but also on image noise. On the other hand, the SUVpeak was a robust metric to image noise level. TOFimproved the variability of SUVmax and SUVpeak. Thus, we were able to harmonize the spatial resolution using SUVpeak based on the phantom study. Because SUVmax was also strongly affected by image noise, sufficient count statistics is essential for SUVmax harmonization. We recommended that TOFreconstruction and SUVpeak metric should be used to harmonize SUV.