Steady advances in radiotherapy treatment modalities simultaneously yield a demand for more accurate dose calculation techniques. Owing to the stochastic nature of particle interactions, transport problems can naturally be addressed by means of Monte Carlo methods. Monte Carlo simulations represent one of the most rigorous modalities to obtain spatial fluence and dose distributions accompanying the penetration of particles into matter. General-purpose Monte Carlo packages have gained an increasing importance in medical physics simulations. One such Monte Carlo code system is Geant4 - a versatile toolkit for simulating the coupled transport of a large variety of particles. The toolkit is developed by an international collaboration under participation of CERN.
This thesis presents a comprehensive and detailed examination of Geant4 physics algorithms pertinent to radiotherapy simulations. Available physics options are partly extended. The accuracy of Monte Carlo models embedded in Geant4 is assessed by means of a series of radiation transport benchmarks covering different aspects of electron and ion transport.
A significant part of the thesis is dedicated to the validation of electron condensed history algorithms. Monte Carlo predictions are systematically benchmarked against experimental data reported in the literature, including dose distributions, backscatter coefficients, energy albedos, as well as angular distributions of electrons backscattered from solid targets. Obtained simulation predictions are evaluated for their ability to describe the variation of experimental data with kinetic energy, angle of incidence, and atomic number of materials. Energies relevant to applications in radiotherapy are covered. Different multiple scattering and energy loss models are compared and their accuracy and limitations are discussed.
Complementing the electron transport studies, the thesis addresses light-ion transport in matter. Physics options in Geant4 are extended by incorporating a parameterization model, based on ICRU 73 stopping powers, to describe the electronic energy loss of ions. With particular attention paid to the recent developments, the accuracy of current Geant4 models is examined for simulating dose profiles of C-12 ions in phantom materials. Obtained distributions are validated against experimental data available in the literature. A quantitative analysis is performed addressing the precision of the Bragg peak position and proportional features of dose distributions. In addition, the effect of different generators for ion fragmentation on dose profiles is evaluated.