Abstract
This thesis investigates the use of Monte Carlo methods to study several imaging applications. Studies are based on the EGS4 code system, and some of these used a low energy electron expansion for this software package that was developed during this work. The code is firstly used to gain some insight into one-sided imaging techniques making use of megavoltage radiation. The method explored is based on induced positron annihilation. The dependence of annihilation yields on atomic number demonstrates that the technique is suitable for the inspection of high-Z inclusions in low-Z, less dense, matrices. Results obtained with the EGS4 code were found to be in good agreement with experimental data. Several applications have been considered via the simulation approach, showing that areas like civil engineering and nuclear material inspection can benefit from this novel inspection technique. The limited accuracy of EGS4 in the simulation of X-ray tubes operating at diagnostic energies led to an expansion of the code to be developed. The enhanced code incorporates a generalized oscillator strength (GOS) model for electron atom inelastic collisions, where atomic bound effects are considered. An enhanced version of this model has been developed so that K-shell ionisation events could be reproduced realistically. The accuracy of low energy bremsstrahlung emissions has also been assessed, and an improved scheme for the angular sampling of newly created photons suggested. The low energy electron expansion of EGS4 (the EGS4/GOS code) is described, and used to simulate photon spectra from diagnostic X-ray tubes. The results were compared with experimental data, showing an accuracy of the order of 15% near the Ka line. Some X-ray tube design studies were carried out using the EGS4/GOS code. The importance of the different physical interactions was analysed. The results show that better fluorescence-to-bremsstrahlung ratios can be obtained with thinner targets, but a factor 2 increase in this ratio is achieved at the expense of a decrease of 40 times in efficiency. The need for low-Z substrates in thin-target applications is also discussed.