Combining Mathematical Programming and Heuristic Methods to Solve the Multi-Objective Beam Angle Optimisation Problem in Radiation Therapy

Abstract

Cancer is an increasingly common disease. Only in New Zealand, more than 20,000 people are diagnosed with cancer each year. According to the Ministry of Health of New Zealand, one in three New Zealanders will have some experience of cancer, either personally or through a relative or friend. Furthermore, cancer is the leading cause of death (28.9 percent) in New Zealand and a major cause of hospitalisation. Although the prevalence of cancer has increased in the last four decades, cancer survival has doubled in the same period. Improvements in cancer survival rates are mainly explained by the great e orts that have been made by researchers in this area, developing new techniques and drugs to ght the disease. One commonly used technique in cancer treatment is radiation therapy. Radiation therapy can be used either as a standalone therapy modality or, more often, in combination with other modalities such as surgery, chemotherapy, or immunotherapy. Radiation therapy aims to deliver the prescribed dose to tumour cells, while surrounding organs receive a dose as low as possible, that is, the aim is to damage tumour cells without negative side e ects. Intensity modulated radiation therapy (IMRT) is the most commonly used technique in radiotherapy since it allows to achieve a good dose coverage while avoiding surrounding organs. The problem of nding a treatment plan that maximises the radiation delivered to the tumour while minimising radiation delivered to the surrounding organs at risk is inherently multi-objective, as the goal of maximising tumor coverage con icts with the goal of minimizing damage to the organs at risk. In spite of that, the vast majority of research e orts made so far have formulated this problem as a single objective one. Only in the last two decades more attention has been paid to multi-objective radiation therapy problems. As a consequence of the lack of research in this area, most commercial software used in clinical practice does still not include multi-objective algorithms to produce treatment plans. This thesis is mainly focused on a particular problem arising in multi-objective IMRT, called the multi-objective beam angle optimisation problem. The aim of this thesis is to provide new strategies to solve this problem from a multi-objective point of view by combining heuristic algorithms and mathematical programming so it can help, in the near future, treatment planners in clinical practice to keep improving the survival rates.