Solving Beam-Angle Selection and Dose Optimization Simultaneously via High-Throughput Computing

Abstract

We provide a framework for integrating two stages of radiation treatment planning (RTP): beam-angle selection (BAS) and dose optimization (DO). The framework is applied to both classical three-dimensional conformal radiotherapy and advanced intensity-modulated radiation therapy. Automated BAS and improved dose distribution are achieved within the framework. A metaheuristic approach, nested partitions, is applied. Alternative BAS and DO algorithms or commercial RTP software and clinical experience can be embedded within the framework to provide new methods for warm starts and evaluations of the quality of beam-angle set samples. Computational efficiency is achieved by utilizing high-throughput computing via the Condor system. Computational results show that our framework has led to a significant improvement in terms of solution quality and delivery time compared with current clinical practice.

Treatment Planning for Volumetric-Modulated Arc Therapy: Model and Heuristic Algorithms

“…Notice that is binary variable, then we prove by two cases. If , from (26) and (27), we obtain , which is equivalent to (24). Similarly, if , from (26) and (28), we obtain , which is again equivalent to (24).…”

“…If , from (26) and (27), we obtain , which is equivalent to (24). Similarly, if , from (26) and (28), we obtain , which is again equivalent to (24). We obtain the linearized model by substituting (2) with (32) and adding (26)-(28).…”

“…NP framework is a new and powerful optimization method that combines intelligent global sampling with local heuristic search. NP-based approaches have been applied in IMRT treatment planning [24] and lot sizing problems [22]. The general idea of NP is to partition the solution region into subregions that can be analyzed and evaluated independently (even in parallel), and compares among all subregions to determine how to guide the partitioning direction.…”

A surrogate-based metaheuristic global search method for beam angle selection in radiation treatment planning

“…Shi and Olafsson (2000) pointed out that the optimal solutions correspond to the set of absorbing states and the Markov chain will eventually reach one of the absorbing states. Finite time behavior of NP was discussed in Shi and Olafsson (2000) and further analyzed for the beam angle selection problem by Zhang et al (2009). In particular, the expected number E[Y] of iterations until the NP Markov chain gets absorbed is given by, $$E\left[Y\right]=\frac{1}{P_0^{d^{\ast }}}true(\frac{true(1-{\left(\frac{1-P_0}{M}\right)}^{d^{\ast }}true)(1-P_0)(M-1)}{P_0(M-1+P_0)}+\frac{P_0-P_0^{d^{\ast }+1}}{1-P_0}true)$$ where P 0 is the success probability of NP, which is the probability of the algorithm moving in the correct direction.…”

A linearized model and nested-partitions heuristics for VMAT radiation treatment planning optimization

“…NP framework is a new and powerful optimization method that combines intelligent global sampling with local heuristic search [11].NP-based approaches have been applied in IMRT treatment planning [12] and lot sizing problems [13]. The general idea of NP is to partition the solution research into subregions that can be analyzed and evaluated independently (even in parallel), and compares among all subregions to determine how to guide the partitioning process.…”