A model for the control of a multileaf collimator in radiation therapy treatment planning

Abstract

A new approach for inverse treatment planning in radiation therapy with the multileaf collimator (MLC) technique is presented. The application of the MLC technique requires an algorithm for the computation of the positions or velocities of leaves as a function of time such that the prescribed dose in the patient space is obtained. The conventional method is to first determine the intensity distribution in the treatment space by applying some inverse treatment techniques. Then an MLC control algorithm which generates the determined intensity distribution is required. In this paper an MLC control algorithm not requiring an intensity distribution as an intermediate step is expressed. The algorithm uses the MLC leaf positions and delivery times (in multiple static treatments) directly in inverse treatment planning. The method is implemented to multiple static treatments but a modification to dynamical MLC techniques is possible. The numerical test is based on the search for a feasible solution but also the optimization (object function) based approach is formulated. The related feasible problem consists of a high-dimensional system of algebraic (in-) equations. It is solved by an iterative row action projection method. The applied algorithm is well suited and computationally robust for high-dimensional feasible problems.