Optimisation approaches for Intensity Modulated Radiation Therapy
This PhD project was carried out in collaboration with RaySearch Laboratories AB.
People
The PhD student was Fredrik Carlsson with Anders Forsgren (CIAM/KTH) as the advisor. In addition to the adviser, the reference group consisted of Johan Löf (RaySearch Laboratories AB), and Henrik Rehbinder (RaySearch Laboratories AB).
Financing
The project was fully funded by Swedish Research Council.
Status
The project was completed in 2008-04-25 when Fredrik Carlsson successfully defended his PhD thesis at the division of Optimization and Systems Theory, Department of Mathematics, KTH School of Engineering Sciences. The project was also selected as a show case project for European success stories in industrial mathematics.
Background
IMRT (Intensity Modulated Radiation Therapy) is an advanced mode of high-precision radiotherapy that utilises computer-controlled linear accelerators to deliver precise radiation doses to a malignant tumour or specific areas within the tumour. IMRT allows for the radiation dose to conform more precisely to the three-dimensional shape of the tumour by modulating the intensity of the radiation beam in multiple small volumes. IMRT also allows higher radiation doses to be focused to regions within the tumour while minimising the dose to surrounding normal critical structures. Due to the complexity of IMRT treatment plans, optimisation methods are needed to design high quality treatments.
Goals
To develop and evaluate optimisation approaches for IMRT with emphasis on numerical efficiency and treatment delivery aspects.
Scientific achievements
The main scientific achievement is to utilise specific properties of the optimisation problem that arises in IMRT. The thesis includes studies of how to explore the inherent ill-conditioning of the pencil-weight optimisation problem by eigenvalue computations, quasi-Newton methods and column-generation techniques. The first two papers deal with strategies for solving fluence map optimisation problems efficiently while avoiding solutions with jagged fluence profiles. The last two papers concern optimisation of step-and-shoot parameters with emphasis on generating treatment plans that can be delivered efficiently and accurately. Numerical results demonstrate that the adjustment of leaf positions improves the plan quality and that satisfactory treatment plans are found with few segments. The method developed provide a tool for exploring the trade-off between plan quality and treatment complexity by generating a sequence of deliverable plans of increasing quality.
