Abstract
PURPOSE: Contemporary PET/CT scanners can use 70-min dynamic whole-body (D-WB) PET to generate more quantitative information about FDG uptake than just the SUV by generating parametric images of FDG metabolic rate (MR(FDG)). The analysis requires the late (50-70 min) D-WB tissue data combined with the full (0-70 min) arterial input function (AIF). Our aim was to assess whether the use of a scaled population-based input function (sPBIF) obviates the need for the early D-WB PET acquisition and allows for a clinically feasible 20-min D-WB PET examination. METHODS: A PBIF was calculated based on AIFs from 20 patients that were D-WB PET scanned for 120 min with simultaneous arterial blood sampling. MR(FDG) imaging using PBIF requires that the area under the curve (AUC) of the sPBIF is equal to the AUC of the individual patient's input function because sPBIF AUC bias translates into MR(FDG) bias. Special patient characteristics could affect the shape of their AIF. Thus, we validated the use of PBIF in 171 patients that were divided into 12 subgroups according to the following characteristics: diabetes, cardiac ejection fraction, blood pressure, weight, eGFR and age. For each patient, the PBIF was scaled to the aorta image-derived input function (IDIF) to calculate a sPBIF, and the AUC bias was calculated. RESULTS: We found excellent agreement between the AIF and IDIF at all times. For the clinical validation, the use of sPBIF led to an acceptable AUC bias of 1-5% in most subgroups except for patients with diabetes or patients with low eGFR, where the biases were marginally higher at 7%. Multiparametric MR(FDG) images based on a short 20-min D-WB PET and sPBIF were visually indistinguishable from images produced by the full 70-min D-WB PET and individual IDIF. CONCLUSIONS: A short 20-min D-WB PET examination using PBIF can be used for multiparametric imaging without compromising the image quality or precision of MR(FDG). The D-WB PET examination may therefore be used in clinical routine for a wide range of patients, potentially allowing for more precise quantification in e.g. treatment response imaging.