Optimisation of Radiation Therapy

Optimisation of Radiation Therapy

The project concerns new optimisation methods for intensity-modulated radiation therapy.

Challenge overview

RaySearch Laboratories, which was founded in 2000 as a spin-off from Karolinska Institutet, is a medical high-tech company with a strong research profile. The CEO and founder has an undergraduate degree from KTH, and he had good connections with the Division of Optimisation and Systems Theory at KTH. The initiative to launch an industrial graduate student project jointly with KTH was made to take a deep look at the mathematical aspects of the optimisation models and optimisation problems that the company faces. The project was financially supported by the Swedish Research Council (VR) and RaySearch.

Implementation of the initiative

The first part of the project was carried out as an industrial graduate student project during 2003- 2008. The setup was such that the student was employed by RaySearch and a graduate student at KTH. Besides his research project and PhD student courses, he also had part-time company duties at RaySearch, making him well acquainted with the academic environment at KTH as well as the industrial environment at RaySearch. A reference group with members from KTH and RaySearch was formed to support the project. The group usually met twice a
year. Informal meetings were held more frequently between advisors from KTH and RaySearch, so as to make the project run smoothly. The Swedish Research Council provided a contract regulating confidentiality issues, which was helpful.

The problem

The optimisation problem arising in intensitymodulated radiation therapy is an inverse problem in that certain requirements on the desired dose distribution in the patient are typically known. These requirements are in general conflicting since high dose is required in tumour cells whereas sensitive organs must be spared. A major challenge is how to formulate the optimisation problem so that the solution obtained is clinically acceptable while it at the same time best fulfils the given requirements. The PhD student project was aimed at utilising problem structure to analyse and design methods for solving the optimisation problem in order to achieve these goals. Results and achievements

The PhD student project led to significantly deepened understanding of the optimisation problems that arise. The interaction between problem structure and behaviour of methods led to important insights into how the problems can be solved efficiently.

Lessons learned and replicability

An important lesson in this project is the interplay between practice and theory. The understanding of the behaviour of methods required a deep understanding of properties of the optimisation problems. Conversely, practical aspects of the optimisation problems led to new interesting fundamental research questions on optimisation methods. Another important lesson is the close interplay between industry and academia which is necessary for such a project to be successful. Trust and openness from both parties is essential for success. The PhD student project discussed above was successfully completed in 2008. It has been succeeded by two new PhD student projects, using the same framework for cooperation between RaySearch and KTH.

Partners in the project

RaySearch Laboratories, Stockholm, Sweden
Johan Löf (johan.lof@raysearchlabs.com)
F. Carlsson (fredrik.carlsson@raysearchlabs.com)

Department of Mathematics – KTH, Stockholm, Sweden
Anders Forsgren (andersf@kth.se)

Center for Industrial and Applied Mathematics (CIAM) at KTH