Validation of a small-animal PET simulation using GAMOS: a GEANT4-based framework.

Monte Carlo-based modelling is a powerful tool to help in the design and optimization of positron emission tomography (PET) systems. The performance of these systems depends on several parameters, such as detector physical characteristics, shielding or electronics, whose effects can be studied on the basis of realistic simulated data. The aim of this paper is to validate a comprehensive study of the Raytest ClearPET small-animal PET scanner using a new Monte Carlo simulation platform which has been developed at CIEMAT (Madrid, Spain), called GAMOS (GEANT4-based Architecture for Medicine-Oriented Simulations). This toolkit, based on the GEANT4 code, was originally designed to cover multiple applications in the field of medical physics from radiotherapy to nuclear medicine, but has since been applied by some of its users in other fields of physics, such as neutron shielding, space physics, high energy physics, etc. Our simulation model includes the relevant characteristics of the ClearPET system, namely, the double layer of scintillator crystals in phoswich configuration, the rotating gantry, the presence of intrinsic radioactivity in the crystals or the storage of single events for an off-line coincidence sorting. Simulated results are contrasted with experimental acquisitions including studies of spatial resolution, sensitivity, scatter fraction and count rates in accordance with the National Electrical Manufacturers Association (NEMA) NU 4-2008 protocol. Spatial resolution results showed a discrepancy between simulated and measured values equal to 8.4% (with a maximum FWHM difference over all measurement directions of 0.5 mm). Sensitivity results differ less than 1% for a 250-750 keV energy window. Simulated and measured count rates agree well within a wide range of activities, including under electronic saturation of the system (the measured peak of total coincidences, for the mouse-sized phantom, was 250.8 kcps reached at 0.95 MBq mL(-1) and the simulated peak was 247.1 kcps at 0.87 MBq mL(-1)). Agreement better than 3% was obtained in the scatter fraction comparison study. We also measured and simulated a mini-Derenzo phantom obtaining images with similar quality using iterative reconstruction methods. We concluded that the overall performance of the simulation showed good agreement with the measured results and validates the GAMOS package for PET applications. Furthermore, its ease of use and flexibility recommends it as an excellent tool to optimize design features or image reconstruction techniques.