Distributed approximation of Pareto surfaces in multicriteria radiation therapy treatment planning
2013 (English)In: Physics in Medicine and Biology, ISSN 0031-9155, E-ISSN 1361-6560, Vol. 58, no 11, 3501-3516 p.Article in journal (Refereed) Published
We consider multicriteria radiation therapy treatment planning by navigationover the Pareto surface, implemented by interpolation between discretetreatment plans. Current state of the art for calculation of a discreterepresentation of the Pareto surface is to sandwich this set between inner andouter approximations that are updated one point at a time. In this paper, wegeneralize this sequential method to an algorithm that permits parallelization.The principle of the generalization is to apply the sequential method to anapproximation of an inexpensive model of the Pareto surface. The informationgathered from the model is sub-sequently used for the calculation of pointsfrom the exact Pareto surface, which are processed in parallel. The model isconstructed according to the current inner and outer approximations, and givena shape that is difﬁcult to approximate, in order to avoid that parts of the Paretosurface are incorrectly disregarded. Approximations of comparable quality tothose generated by the sequential method are demonstrated when the degree ofparallelization is up to twice the number of dimensions of the objective space.For practical applications, the number of dimensions is typically at least ﬁve,so that a speed-up of one order of magnitude is obtained.
Place, publisher, year, edition, pages
2013. Vol. 58, no 11, 3501-3516 p.
IdentifiersURN: urn:nbn:se:kth:diva-122666DOI: 10.1088/0031-9155/58/11/3501ISI: 000318966200004ScopusID: 2-s2.0-84878249203OAI: oai:DiVA.org:kth-122666DiVA: diva2:623182
QC 201305272013-05-242013-05-242013-06-27Bibliographically approved