Unified description and validation of Monte Carlo simulators in PET

Publisher Website
Google Scholar
Abstract
Several Monte Carlo simulators are currently available for positron emission tomography (PET). Because each code has been described in a different way, it is difficult to know which one is best suited to a specific application. To help clarify the capabilities and accuracy of different codes dedicated to PET simulations, we propose a uniform description of the code features. This description specifies features pertaining to the models used for simulating the physics of PET and for describing a PET acquisition, to the acceleration strategies and to the technical characteristics of the code implementation. To assess the code accuracy, we suggest validation procedures based on NEMA phantoms involving standard physical parameters and simulation of a complex activity distribution. A test characterizing the statistical properties of detected coincidences is also described. The proposed code description and validation procedures are illustrated by considering the SimSET and PET-EGS codes. These codes differ in many features, including models for randoms and dead time, and source description. Despite these differences, both codes yielded data with properties close to those of real data. Depending on the intended application, one code might be preferred however. Indeed, only PET-EGS allows for accurate modelling of count rates while SimSET is more computationally efficient. The proposed code description and validation procedures might help determine which code is most appropriate for a specific application.