Zhu Lei, Niu Tianye, Choi Kihwan, Xing Lei
George W. Woodruff School, Nuclear and Radiological Engineering and Medical Physics Programs, Georgia Institute of Technology, Atlanta, Georgia 30332, USA.
Technol Cancer Res Treat. 2012 Apr;11(2):149-62. doi: 10.7785/tcrt.2012.500244.
Intensity modulated arc therapy (IMAT) delivers conformal dose distributions through continuous gantry rotation with constant or variable speed while modulating the field aperture shape and weight. The enlarged angular space and machine delivery constraints make inverse planning of IMAT more intractable as compared to its counterpart of fixed gantry IMRT. Currently, IMAT inverse planning is being done using two extreme methods: the first one computes in beamlet domain with a subsequent arc leaf sequencing, and the second proceeds in machine parameter domain with entire emphasis placed on a pre-determined delivery method without exploring potentially better alternative delivery schemes. Towards truly optimizing the IMAT treatment on a patient specific basis, in this work we propose a total-variation based inverse planning framework for IMAT, which takes advantage of the useful features of the above two existing approaches while avoiding their shortcomings. A quadratic optimization algorithm has been implemented to demonstrate the performance and advantage of the proposed approach. Applications of the technique to a prostate case and a head and neck case indicate that the algorithm is capable of generating IMAT plans with patient specific numbers of arcs efficiently. Superior dose distributions and delivery time are achieved with a maximum number of apertures of three for each field. As compared to conventional beamlet-based algorithms, our method regularizes the field modulation complexity during optimization, and permits us to obtain the best possible plan with a pre-set modulation complexity of fluences. As illustrated in both prostate and head-and-neck case studies, the proposed method produces more favorable dose distributions than the segment-based algorithms, by optimally accommodating the clinical need of intensity modulation levels for each individual field. On a more fundamental level, our formulation preserves the convexity of optimization and makes the search of the global optimal solution possible with a deterministic method.
调强弧形治疗（IMAT）通过机架以恒定或可变速度连续旋转同时调制射野孔径形状和权重来实现适形剂量分布。与固定机架调强放疗相比，IMAT增大的角度空间和机器输出限制使得其逆向计划更加棘手。目前，IMAT逆向计划采用两种极端方法：第一种在子野域进行计算，随后进行弧形准直器叶片排序；第二种在机器参数域进行，完全侧重于预先确定的输出方法，而不探索潜在更好的替代输出方案。为了在患者个体基础上真正优化IMAT治疗，在这项工作中，我们提出了一种基于全变差的IMAT逆向计划框架，该框架利用了上述两种现有方法的有用特征，同时避免了它们的缺点。已实现一种二次优化算法来证明所提方法的性能和优势。将该技术应用于一个前列腺病例和一个头颈部病例表明，该算法能够有效地生成具有患者个体特定弧数的IMAT计划。每个射野最多三个孔径时可实现更好的剂量分布和输出时间。与传统的基于子野的算法相比，我们的方法在优化过程中规范了射野调制复杂性，并允许我们在预设的注量调制复杂性下获得尽可能好的计划。如前列腺和头颈部病例研究所示，所提方法通过最佳地适应每个个体射野的调强水平的临床需求，产生比基于段的算法更有利的剂量分布。在更基本的层面上，我们的公式保持了优化的凸性，并使得用确定性方法搜索全局最优解成为可能。
