a comparison between gate and mcnpx monte carlo codes in simulation of medical linear accelerator

radiotherapy dose calculations can be evaluated by monte carlo (mc) simulations with acceptable accuracy for dose prediction in complicated treatment plans. in this work, standard, livermore and penelope electromagnetic (em) physics packages of geant4 application for tomographic emission (gate) 6.1 were compared versus monte carlo n-particle extended (mcnpx) 2.6 in simulation of 6 mv photon linac. to do this, similar geometry was used for the two codes. the reference values of percentage depth dose (pdd) and beam profiles were obtained using a 6 mv elekta compact linear accelerator, scanditronix water phantom and diode detectors. no significant deviations were found in pdd, dose profile, energy spectrum, radial mean energy and photon radial distribution, which were calculated by standard and livermore em models and mcnpx, respectively. nevertheless, the penelope model showed an extreme difference. statistical uncertainty in all the simulations was <1%, namely 0.51%, 0.27%, 0.27% and 0.29% for pdds of 10 cm 2 × 10 cm 2 filed size, for mcnpx, standard, livermore and penelope models, respectively. differences between spectra in various regions, in radial mean energy and in photon radial distribution were due to different cross section and stopping power data and not the same simulation of physics processes of mcnpx and three em models. for example, in the standard model, the photoelectron direction was sampled from the gavrila-sauter distribution, but the photoelectron moved in the same direction of the incident photons in the photoelectric process of livermore and penelope models. using the same primary electron beam, the standard and livermore em models of gate and mcnpx showed similar output, but re-tuning of primary electron beam is needed for the penelope model.