A two-stage approach for VMAT treatment plan optimization

In this paper we study the optimization problem for the radiation treatment planning of Volumetric-Modulated Arc Therapy (VMAT). Considering dose rate and MLC leafs as decision variables, we formulate the problem as nonlinear integer programming problem, which is very difficult to solve. A two-stage approach is proposed to tackle it, where in the first stage integer decision variables for MLC leafs are relaxed so as to decompose the nonlinear optimization problem using generalized Bender's decomposition, and in the second stage an 0-1 mixed integer programming problem is solved by nested-partitions method for MLC leafs positions. Throughout this approach a number of linear programming (LP) problems are solved, hence a high quality feasible treatment plan can be efficiently obtained.