Utilizing the Degrees of Freedom in Radiation Therapy Optimization

Abstract: The development of advanced software for radiation therapy treatment has had a significant role in the last decades’ improvements in cancer treatment. By optimizing the radiation dose given to each patient, the probability of a successful treatment increases. The development of mathematical optimization methods, required to produce as good radiation therapy treatment plans as possible, is therefore of high importance. When producing treatment plans for Volumetric Modulated Arc Therapy (VMAT) today, there are degrees of freedom that are not fully utilized. The possibility of rotating the patient couch and collimator during radiation delivery, to achieve a better dose distribution over the patient, is not taken advantage of. In order to take full advantage of these extra degrees of freedom, methods for predicting how good different arcs will perform during VMAT-delivery need to be developed. Since the number of possible VMAT-arcs is very large, such a method needs to be very time efficient in order to be of conventional use. In this thesis, a heuristic method for finding good VMAT-arcs for patients with multiple brain metastasis is proposed. In the method, the sphere around the patient’s head is discretized into a number of beam directions. From each of these beam directions a projection of the patient’s anatomy is produced, and based on the relative location of the cancer metastasis and the organs desired to protect, a cost is assigned. By linking together adjacent beam direction, VMAT-arcs can be created, and each possible VMATarc is assigned an arc cost depending on the cost of the beam directions it traverses. Such arc cost is used as an indication of how good a VMAT-arc will perform during VMAT-delivery. The heuristic method is tested and evaluated on four different patient cases. In three of the patient cases, the proposed method gives good results. Overall the results indicate that it is possible to predict how good a VMAT-arc will perform during VMAT-delivery, by looking at the relative locations of the target and organ projections in a discrete number of points along the arc. However, since the number of test cases are few, no statistically significant conclusions can be drawn.