Zinchenko Y, Craig T, Keller H, Terlaky T, Sharpe M
Advanced Optimization Laboratory, Department of Computing and Software, McMaster University, Hamilton L8S 4K1, Canada.
Phys Med Biol. 2008 Jun 21;53(12):3231-50. doi: 10.1088/0031-9155/53/12/011. Epub 2008 May 27.
Radiation therapy is an important modality in treating various cancers. Various treatment planning and delivery technologies have emerged to support intensity modulated radiation therapy (IMRT), creating significant opportunities to advance this type of treatment. However, one of the fundamental questions in treatment planning and optimization, 'can we produce better treatment plans relying on the existing delivery technology?' still remains unanswered, in large part due to the underlying computational complexity of the problem, which, in turn, often stems from the optimization model being non-convex. We investigate the possibility of including the dose prescription, specified by the dose-volume histogram (DVH), within the convex optimization framework for inverse radiotherapy treatment planning. Specifically, we study the quality of approximating a given DVH with a superset of generalized equivalent uniform dose (gEUD)-based constraints, the so-called generalized moment constraints (GMCs). As a bi-product, we establish an analytic relationship between a DVH and a sequence of gEUD values. The newly proposed approach is promising as demonstrated by the computational study where the rectum DVH is considered. Unlike the precise partial-volume constraints formulation, which is commonly based on the mixed-integer model and necessitates the use of expensive computing resources to be solved to global optimality, our convex optimization approach is expected to be feasible for implementation on a conventional treatment planning station.
放射治疗是治疗各种癌症的重要手段。各种治疗计划和实施技术不断涌现，以支持调强放射治疗（IMRT），为推进此类治疗创造了重大机遇。然而，治疗计划与优化中的一个基本问题，即“依靠现有的实施技术，我们能否制定出更好的治疗计划？”，在很大程度上仍未得到解答，这主要是由于该问题潜在的计算复杂性，而这又往往源于优化模型的非凸性。我们研究了在逆向放射治疗计划的凸优化框架内纳入由剂量体积直方图（DVH）指定的剂量处方的可能性。具体而言，我们研究了用基于广义等效均匀剂量（gEUD）的约束（即所谓的广义矩约束（GMC））的超集来近似给定DVH的质量。作为一个副产品，我们建立了DVH与一系列gEUD值之间的解析关系。如在考虑直肠DVH的计算研究中所表明的，新提出的方法很有前景。与通常基于混合整数模型且需要使用昂贵计算资源才能求解到全局最优的精确部分体积约束公式不同，我们的凸优化方法有望在传统治疗计划工作站上可行地实施。
