A Global Score-Driven Beam Angle Optimization in IMRT

Abstract

Abstract. Radiation therapy is one of the main treatment modalities for cancer. The objective of radiation therapy is to eliminate all cancer cells by delivering a prescribed dose of radiation to the tumor volume while sparing at the same time the surrounding tissues. Intensitymodulated radiation therapy (IMRT) is a sophisticated technologicallydriven type of radiation therapy where non-uniform radiation fields are used to irradiate the patient from different beam angle directions. Appropriate selection of beam irradiation directions -beam angle optimization (BAO) problem -enhance the quality of the treatment plan. The BAO problem is a very difficult global non-convex optimization problem for which there are few or none commercial solutions. Typically, the BAO procedure is driven by the outcome of the fluence map optimization (FMO) problem -the problem of calculating the most adequate radiation intensities. However, functions used for modeling the FMO problem have little clinical meaning. Typically, selection/validation of treatment plans is done considering a set of dosimetric measures. In this study, we propose a treatment plan global score, based on dosimetric criteria and its relative importance, as alternative plan's quality measure to drive the BAO procedure. For the clinical case of nasopharyngeal tumor, the use of a global score to drive the BAO procedure lead to higher quality treatment plans. For similar target coverage, an improved organ sparing was obtained.

Beam angle optimization in IMRT: are we really optimizing what matters?

“…In this paper, a new score based on clinical dosimetric values weighted by its clinical importance, as described by Ventura et al (2016), is introduced and embedded in a parallel multistart derivativefree optimization framework that explores thoroughly the BAO problem continuous search space. Preliminary tests using this score to guide a simpler BAO search lead to promising results for a singlepatient case (Rocha et al, 2017). Unlike previous scores proposed for BAO, this score is obtained after solving the FMO problem and thus BAO and FMO are considered simultaneously.…”