Noncoplanar beam angle optimization in IMRT treatment planning using pattern search methods

Humberto Rocha,Joana M Dias,Brígida C Ferreira,Maria C Lopes

doi:10.1088/1742-6596/616/1/012014

Abstract

Radiation therapy is used to treat localized cancers, aiming to deliver a dose of radiation to the tumor volume to sterilize all cancer cells while minimizing the collateral effects on the surrounding healthy organs and tissues. The planning of radiation therapy treatments requires decisions regarding the angles used for radiation incidence, the fluence intensities and, if multileaf collimators are used, the definition of the leaf sequencing. The beam angle optimization problem consists in finding the optimal number and incidence directions of the irradiation beams. The selection of appropriate radiation incidence directions is important for the quality of the treatment. However, the possibility of improving the quality of treatment plans by an optimized selection of the beam incidences is seldom done in the clinical practice. Adding the possibility for noncoplanar incidences is even more rarely used. Nevertheless, the advantage of noncoplanar beams is well known. The optimization of noncoplanar beam incidences may further allow the reduction of the number of beams needed to reach a clinically acceptable plan. In this paper we present the benefits of using pattern search methods for the optimization of the highly non-convex noncoplanar beam angle optimization problem.

Highlights

Cancer is one of the most significant health problems worldwide with respect to its incidence and mortality alike
In this paper we present the benefits of using pattern search methods for the optimization of the highly non-convex noncoplanar beam angle optimization (BAO) problem
The poll step, where convergence to a local minima is assured, and the search step, where flexibility is conferred to the method since any strategy can be applied

Summary

Introduction

Cancer is one of the most significant health problems worldwide with respect to its incidence and mortality alike. In most of the previous works on BAO, the entire range, [0◦, 360◦] in the coplanar case, of gantry angles is discretized into spaced beam directions with a given angle increment, such as 5 or 10 degrees, where exhaustive searches are performed directly or guided by a variety of different heuristics including simulated annealing [7], genetic algorithms [19], particle swarm optimization [24] or other heuristics incorporating a priori knowledge of the problem [20] Those global heuristics can theoretically avoid local optima, globally optimal or even clinically better solutions cannot be obtained without a large number of objective function evaluations. Our approach for modeling the noncoplanar BAO problem uses the optimal solution value of the FMO problem as the measure of the quality for a given beam angle set. The FMO model is used as a black-box function and the conclusions drawn regarding BAO coupled with this formulation/resolution are valid if different FMO formulations/resolutions are considered

Pattern Search Methods

Findings

Discussion and conclusions