Accelerated IMRT plan optimization using the adaptive clustering method

• DOI: 10.1016/j.laa.2007.03.025
• Publication Date: 2007-04-09T13:03:36Z
• Authors: Alexander Scherrer; Karl-Heinz Küfer
• Keywords: Numerical Analysis; 03 medical and health sciences; Algebra and Number Theory; 0302 clinical medicine; Intensity-modulated radiation therapy (IMRT) plan optimization; Agglomerative hierarchical clustering methods; Discrete Mathematics and Combinatorics; Geometry and Topology; Sensitivity analysis of nonlinear optimization problems; Aggregation methods in large scale optimization

AbstractA major problem of intensity modulated radiation therapy (IMRT) planning is the high numerical complexity of the corresponding optimization problems. This is mainly a consequence of the volume discretization into a very large number of volume elements. Many approaches simplify the volume discretization in order to reduce the computation times. However, most of them do the simplification in a non-adaptive way prior to the plan computations and are thus of a heuristic type. The adaptive clustering method overcomes this drawback as follows. The dose mapping and the planning structures first undergo a preprocessing. The obtained output then serves as a ‘construction kit’ for the formation of volume discretizations that are individually adapted during each plan computation. This method is highly flexible, since it is widely independent of the type of plan parameters, can be applied to a very large class of functions for dose evaluation, and is not restricted to the use of a particular numerical solver.