Harmonization of standardized uptake values between two scanners, considering repeatability and magnitude of the values in clinical fluorine-18-fluorodeoxyglucose PET settings: a phantom study.

Abstract

The standardized uptake value (SUV) is a quantitative imaging biomarker which has been used the most widely in the field of PET. Current harmonization strategies commonly include image blurring by application of an additional extent of post-reconstruction Gaussian filtering in the aim of recovery coefficient adjustment. The objective of this study was to harmonize SUVmax and SUVpeak measured by an older PET/computed tomography scanner with a newer scanner without depending on image blurring. We performed multiple phantom studies using a count-based method, in which SUVs can be calculated in a repeatable manner. We measured SUVmax and SUVpeak for spheres of various sizes (10, 13, 18, 22, 28, and 37-mm diameter) and actual SUVs (2, 4, 8, and 16) under standard clinical settings. The measured SUVs were compared with the actual SUVs, and the approximate equations were obtained. The relationships of the equations between two different PET/computed tomography scanners, Aquiduo and Celesteion, were examined to obtain conversion equations. The SUVmax and SUVpeak measured by Aquiduo were lower than those measured by Celesteion. For both scanners, the relationships between the actual and measured SUVs were linearly fit (R>0.99) for each sphere size. The equations varied depending on the sphere size. Also, for large spheres (>2 cm), SUVmax and SUVpeak could be harmonized by simple multiplication, regarding the small absolute values of y-intercept (<0.2) as negligible. Size-specific equations to convert SUVmax and SUVpeak measured by Aquiduo into the estimated values on Celesteion were obtained. Also, for large lesions (>2 cm), SUVs may be converted by simple multiplication. These methods might enable simple SUV harmonization which is not based on post-reconstruction Gaussian filtering.