Experimental validation of Monte Carlo simulation model for proton range verification using an in-beam dual-head PET system

Abstract

Proton therapy offers highly localized dose distributions, but its clinical potential is limited by uncertainties in proton range. In-beam positron emission tomography (PET) imaging of proton-induced β+-emitting isotopes provides a promising solution for real-time range verification. This study presents a comprehensive Monte Carlo (MC) simulation model for dual-head PET (DHPET)-based range verification using the GATE platform, with experimental validation performed at Kaohsiung Chang Gung Memorial Hospital, Taiwan. Measurements were conducted using homogeneous phantoms irradiated by proton beams at various energies, and PET images were acquired with a Sumitomo DHPET system. The simulation workflow was divided into two stages: proton-induced β+ isotope production and PET image acquisition. Three nuclear models were implemented and compared: GEANT4's built-in theoretical model (QGSP_BIC), EXFOR-based cross sections, and the latest Nuclear Data Sheets (NDS) dataset. The MC model was validated against RayStation-calculated dose distributions and PET measurements from both point and flood sources. Simulated dose distributions showed excellent agreement with RayStation, with a mean range deviation of ±0.2 mm and detector response deviation within ±0.5%. For PET activity profiles, the NDS dataset achieved the closest match to experimental data, EXFOR showed moderate agreement, while QGSP_BIC, despite discrepancies in activity shape, provided superior distal activity range estimates in gel-water phantoms. Under optimal conditions, proton range verification accuracy using the proposed MC simulation model and updated nuclear cross-section models could reach sub-millimeter precision (within ±1 mm). This work establishes a validated MC framework for in-beam DHPET-based range verification and emphasizes the pivotal role of nuclear cross-section selection in optimizing PET-based range estimation.