Testing and Tuning of Optimization Algorithms : On the implementation of Radiotherapy

Abstract: When treating cancer patients using radiotherapy, careful planning is essential to ensure that the tumour region is treated while surrounding healthy tissue is not injured in the process. The radiation dose in the tumour along with the dose limitations to healthy tissue can be expressed as a constrained optimization problem. The goal of this project has been to create prototype environments in C++ for both testing and parameter tuning of optimization algorithms intended to solve radiotherapy problems. A library of test problems has been implemented on which the optimization algorithms can be tested. For the sake of simplicity, the problem solving and parameter tuning has only been carried out with the interior point solver IPOPT. The results of a parameter tuning process are displayed in tables where the effect of the tuning can be analysed. By using the implemented parameter tuning process, some settings have been found that are better than the default values when solving the implemented test problems.