Print Email Facebook Twitter Local Pareto navigation for treatment plan trade-off analysis in radiotherapy Title Local Pareto navigation for treatment plan trade-off analysis in radiotherapy Author Slotboom, C. Contributor Keijzer, M. (mentor) van Haveren, R. (mentor) Faculty Electrical Engineering, Mathematics and Computer Science Department Delft Institute of Applied Mathematics Programme Applied Mathematics Date 2017-01-02 Abstract The method Erasmus-iCycle is able to generate a treatment plan for cancer patients, based on the patient's anatomy and a predefined wish list. This wish list contains goals for different organs near the tumour, listed by priority. However, it is not certain that the generated iCycle plan is the best fit for an individual patient, since every patient is anatomically unique. The goal of this report is therefore to develop a method that generates alternative treatment plans, from which a physician can choose the most desirable plan for his/her patient. To find useful alternative plans, Pareto optimal plans that are close to the iCycle plan have to be found. This can be done by applying the reference point method (RPM) to different reference paths, which leads to different solutions on the Pareto front. Preferably, these solutions are spread around the iCycle plan in all directions. Besides that, the density of plans has to be high near the iCycle plan and really low on the end parts of the Pareto front, since those solutions are usually not clinically desirable. In the TSRPM, which is the final method of this report, reference paths are lines through the ideal point and one other reference point (for which the coordinates are higher than those of the ideal point, so the reference path is increasing). The reference points are chosen in the tangent space to the Pareto front in the iCycle point. Using reference points spread through the right part of the tangent space, ensures that for each objective, solutions with higher and with lower doses in the corresponding organ are found. By choosing the desired number of reference points and smart boundaries, multiple optimal plans are generated. These plans can be analysed and compared, so the physician can choose a plan. To reference this document use: http://resolver.tudelft.nl/uuid:b5e59708-f3b4-4615-9c55-fae7909e9ecd Embargo date 2020-01-02 Part of collection Student theses Document type bachelor thesis Rights (c) 2017 Slotboom, C. Files PDF BEP Cynthia Slotboom - Lo ... herapy.pdf 1.91 MB Close viewer /islandora/object/uuid:b5e59708-f3b4-4615-9c55-fae7909e9ecd/datastream/OBJ/view