Parameterised Geant4 simulation for total body PET research

Total-body positron emission tomography (PET) imaging has the potential to transform medical care of a number of diseases and augment our knowledge of systems biology. Various detector designs and geometries are currently under development for total-body PET imaging of humans. This variety, in particular the variation in axial field-of-view (aFOV), motivates a need to compare the performance of these devices in a consistent simulated environment. We present an open-source Geant4 simulation package that allows variation of relevant parameters such as the detector aFOV and the tracer radioisotope from the command line. Two simplified detector geometries based on the Siemens Biograph Vision Quadra and United Imaging uEXPLORER models are supported with variable granularity. The intrinsic radioactivity of the detector crystals is fully simulated. The simulation can be viewed with the built-in GUI, and the results are saved in a plain text format for easy analysis. Example Python analysis code is provided with the simulation, demonstrating calculation of the noise equivalent count rate (NECR) figure of merit using an approximation to the NEMA NU 2-2012 standard method. A good agreement between the simulated count rate performance and experimental data is observed for both geometries. The differences in results are attributed to simplifications in the simulation code, namely not accounting for the light-collection efficiency or readout dead-time. We demonstrate the importance of assessing the scanner performance using appropriate phantom length which significantly affects the obtained results. A dependence between the detector aFOV and the length of the source, with peak NECR plateauing as the detector extends beyond the region of interest is also presented.