Monte-Carlo computations of dose distributions for magnetic resonance guided radiotherapy

The goal with radiotherapy is to damage tumour cells with ionizing radiation while minimizing radiation to healthy tissue. In recent years magnetic resonance guided radiotherapy has become increasingly popular. With this technique a magnetic resonance image of the patient anatomy is obtained in connection with the treatment session. The image can be used to adapt the treatment plan for daily changes in the patient anatomy. Magnetic resonance images have superior soft tissue contrast, and delivers no undesired radiation through the imaging process, compared to competitive methods. The strong magnetic field that enables magnetic resonance imaging will alter the motion of electrons that are created in the ionization process, this will in turn affect the dose to the patient. This thesis investigates how the Monte-Carlo dose computation algorithm in a commercial treatment planning system can be adjusted to enable dose computation under the effect of an external magnetic field. The effect on the dose distributions from the magnetic field is investigated and the computed doses are compared against a different software, EGSnrc, which is considered the gold standard. Furthermore, a new numerical integration method to compute the change in direction of electrons due to the magnetic field is developed, with the purpose of reducing the computation time while maintaining accuracy of the computations. A method for how to simplify the transport of electrons outside the patient is also investigated with the purpose of reducing computation time. The comparisons of the computed doses showed strong agreement with EGSnrc. Most dose distributions had a gamma passing rate which was higher than 98% using a 1%/1mm global gamma criterion. The main source of difference was the dose in air and in regions of high dose gradient. The time required for the computations increased with about 50 - 100% compared to the corresponding computations without a magnetic field, the time was about 3 times less than the initial implementation of the algorithm. Furthermore, the simplified algorithm for transporting electrons outside the patient reduced the total time required for the dose computations by a factor 2-5 depending on the situation, while still enabling an accurate computation of dose from electrons that could return to the patient