Monte Carlo Modelling of Kilovoltage Radiotherapy

Kilovoltage radiotherapy is often used to treat inoperable skin cancers and some other benign conditions. The current treatment planning approach in many radiotherapy centres involves manual point dose calculations in a rectangular water geometry. This approach ignores the dosimetric effects of surface irregularities, underlying tissue heterogeneity, and the use of metal shielding. Furthermore, beam collimation using custom lead cutouts is approximated using equivalent circular beam shapes. In the context of radiobiology, previous studies suggest an increased radiobiological effectiveness of kilovoltage energy photon beams compared to their megavoltage counterparts, but this is not accounted for in the manual calculations. While this current treatment planning approach is appropriate for simple treatment cases, it breaks down for more complex setups. It becomes impractical to perform extensive dosimetric evaluations of treatment plans and near-impossible to perform a 3D dose distribution assessment to compare treatment modalities or, in cases of patient re-irradiation, evaluate the dose overlap from a biological point of view. This thesis aims to address some of these limitations, bringing the current simplistic kilovoltage treatment planning approach in line with modern dose calculation techniques. In this work, EGSnrc general purpose Monte Carlo code was used to create a model of the Xstrahl-200 kilovoltage treatment unit. All clinically usable energy-applicator setup combinations were modelled and validated against measurements of different radiation beam characteristics. Agreement to within 3% was achieved for percentage depth dose curves and profiles in water. Half-value layers agreed to within 0.5 mm of absorbed material. Backscatter factors and output factors were within 2% for the majority of setups. A Python-based treatment planning system was developed to aid in simulation setup and result analysis using patient CT images. End-to-end testing of the proposed workflow was performed. Gamma analysis using 3%/2 mm criteria displayed a passing rate of >85% for challenging geometries. The radiobiological effectiveness of the kilovoltage photon beams was investigated using TOPAS-nBio Monte Carlo code, with the derived relative biological effectiveness factor of 1.14 being in agreement with previous estimates. To explore the applications of this system, several clinical case studies were carried out, highlighting the educational benefit of the Python toolkit developed in this work. Applications included quantitative evaluation of kilovoltage dose distributions as well as comparison and summation of dose distributions from different treatment modalities. A reference digital anatomical phantom was developed allowing for simulations to be performed without the need for individual patient CT images. Three years of clinical records were reviewed to identify the most commonly treated sites and setups. A library of treatment cases was developed for educational purposes of clinical staff involved in kilovoltage radiotherapy. Proposed future work addressing i) the limitations identified during this thesis to include more detailed dosimetric investigations in the kV setting, ii) the speed of the workflow to obtain 3D dose distributions of a kV beam on a patient CT and iii) the medical device regulations to potentially use the toolkit developed in this work to plan patient's kV treatments, has been identified.