Absorbed dose estimation for 18F-FDG PET studies: validation of an SUV-based method for different uptake times.

Objectives In a previous study^1,2, an initial evaluation of the ratio of actual-to-reference SUV [SUVR] from FDG, PET images appeared to successfully improve traditional FDG dosimetry by accounting for the biodistribution of FDG shown in PET images. Here, we investigate the impact the FDG uptake times at imaging on the validity SUVR method.

Methods FDG kinetics were modeled using a 3-compartment model and time activity curves created for different values of the model's parameters to represent normal and abnormal FDG uptake. The organ accumulated activity was simulated at 60 and 120 min. The anatomic model included three spheres to represent organs of different size and placed at different distances within a torso-sized compartment with activity equal to the blood pool activity, Geant4 Monte Carlo simulations provided the true organ dose for such "anatomy". Corresponding SUVs at 60 and 120min were calculated.

Results Four realizations of the model and corresponding simulations for dose were carried out. One for normal biodistribution and three various levels of abnormal uptake. Using the SUVR, the dose estimates where within 3.5% of the true dose for the normal biodistribution. For the abnormal uptake the % improvement over the traditional method was 79.8%, 71.2% and 45.15%. The % avg error was <20% for both 60 and 120min uptake. Error with no compensation for actual SUV values was on the average >50% and as large as 112%

Conclusions The results confirm that (1) the traditional estimation of dosimetry, i.e. assign average values from the literature - can differ significantly from the true dosimetry and (2) that even with static images at 60 and 120min, the SUVR method can provide accurate individualized estimates.

Research Support NIH Summer Internship Program and the Intramural Research Program of the NIH Clinical Center