Simulation of Clinical PET Studies for the Assessment of Quantification Methods

Abstract

On this PhD thesis we developed a methodology for evaluating the robustness of SUV measurements based on MC simulations and the generation of novel databases of simulated studies based on digital anthropomorphic phantoms. This methodology has been applied to different problems related to quantification that were not previously addressed. Two methods for estimating the extravasated dose were proposed andvalidated in different scenarios using MC simulations. We studied the impact of noise and low counting in the accuracy and repeatability of three commonly used SUV metrics (SUVmax, SUVmean and SUV50). The same model was used to study the effect of physiological muscular uptake variations on the quantification of FDG-PET studies. Finally, our MC models were applied to simulate 18F-fluorocholine (FCH) studies. The aim was to study the effect of spill-in counts from neighbouring regions on the quantification of small regions close to high activity extended sources.