Experimental validation of monte carlo pencil beam scanning model in heterogeneous media for proton therapy

Sorriaux, Jefferson [UCL] Rossomme, Séverine [UCL] Lee, John Aldo [UCL] Bertrand, Damien [IBA] Vynckier, Stefaan [UCL] Sterpin, Edmond [UCL]

Purpose/Objective To validate experimentally a GATE/GEANT4-based(G4) Monte Carlo (MC) model in heterogeneous media for dedicated pencil beam scanning in proton therapy. Comparisons between measurements and MC simulations using G4 and PENELOPE-proton are presented. A comparison against analytical modeling from commercial TPS is also investigated. This work evaluates the impact of heterogeneities on range prediction, beam shape and depth dose changes. Materials and Methods The MC model for pencil beam based on G4 has been validated in water and PMMA phantoms (Grevillot et al Phys. Med. Biol.(2011)) reproducing pristine Bragg peaks for a series of individual energies (from 100 to 226.7 MeV) with 0.7 mm range and 0.2 mm spot size accuracy. The same optical model was implemented in PENELOPE-proton. In order to validate the beam model in heterogeneous media, phantoms made of stacked slabs with different densities and known compositions were used. Two experimental test cases including solid water (SW), lung (LN-300) and bone (SB3) tissue-equivalent material were investigated. Depth-dose distributions for a monoenergetic single spot and 10x10cm² composite fields were measured using Gafchromic EBT3 films and the ionization chamber (IC) PPC05 in all configurations. To measure accurately the Bragg peak position, a stack of films of 2x2cm² was inserted in the last centimeter of the proton range. Results Figure 1 shows results for one heterogeneous configuration. All doses-to-medium were converted to dose to water using stopping power ratios. Dose distributions were arbitrarily normalized in the middle of the second SW region. Bragg peak positions are reproduced by MC simulations within 1mm in both configurations (Table 1). IC measurements, G4 (binary-cascade) and PENELOPE-proton simulations are within 2%/2 mm. Point-to-point mean difference of 1.2% is observed between G4 (precompound) and measurement in the first 15 cm of the phantom and increased to 7.2% after bone insert until 286.5mm depth. In lung and bone slabs, EBT3 films and G4 binary-cascade are inagreement within 0.77% while a mean point-to-point difference up to 1.26% is observed with G4 precompound model. The uncertainty (1σ) on EBT3 films was evaluated to be at 2.75% which included readout process and dose calibration against IC (TRS-398). A statistical uncertainty of 0.1% was achieved for MC simulations. Conclusions The Bragg peak position is predicted with 1mm precision for all MC simulations, even though ionization potential values for phantom slabs were calculated using classical additive rules. G4/GATE beam model reproduce depth-dose behavior of proton transport regarding both IC and EBT3 measurement in heterogeneities.