An experimental approach to improve the Monte Carlo modelling of offline PET/CT-imaging of positron emitters induced by scanned proton beams

We report on the experimental campaign carried out at the Heidelberg Ion-Beam Therapy Center (HIT) to optimize the Monte Carlo (MC) modelling of proton-induced positron-emitter production. The presented experimental strategy constitutes a pragmatic inverse approach to overcome the known uncertainties in the modelling of positron-emitter production due to the lack of reliable cross-section data for the relevant therapeutic energy range. This work is motivated by the clinical implementation of offline PET/CT-based treatment verification at our facility. Here, the irradiation induced tissue activation in the patient is monitored shortly after the treatment delivery by means of a commercial PET/CT scanner and compared to a MC simulated activity expectation, derived under the assumption of a correct treatment delivery. At HIT, the MC particle transport and interaction code FLUKA is used for the simulation of the expected positron-emitter yield. For this particular application, the code is coupled to externally provided cross-section data of several proton-induced reactions. Studying experimentally the positron-emitting radionuclide yield in homogeneous phantoms provides access to the fundamental production channels. Therefore, five different materials have been irradiated by monoenergetic proton pencil beams at various energies and the induced beta(+) activity subsequently acquired with a commercial full-ring PET/CT scanner. With the analysis of dynamically reconstructed PET images, we are able to determine separately the spatial distribution of different radionuclide concentrations at the starting time of the PET scan. The laterally integrated radionuclide yields in depth are used to tune the input cross-section data such that the impact of both the physical production and the imaging process on the various positron-emitter yields is reproduced. The resulting cross-section data sets allow to model the absolute level of measured beta(+) activity induced in the investigated targets within a few per cent. Moreover, the simulated distal activity fall-off positions, representing the central quantity for treatment monitoring in terms of beam range verification, are found to agree within 0.6 mm with the measurements at different initial beam energies in both homogeneous and heterogeneous targets.