无商业证书环境下的跨中心食管癌自动计划实现

目的:RapidPlan自动计划可以带来效率、质量和一致性的提升，但对Eclipse计划系统版本和商业证书（license）的较高要求限制了其在资源相对匮乏的基层医院应用。本工作以食管癌为例，探究在没有商业证书和较低版本计划系统环境下，跨中心、跨技术远程应用RapidPlan模型进行自动优化的可行性和剂量学表现，在不增加医院成本的情况下使更多患者受益。方法:（1）选取绵阳市第三人民医院的26例不同分段食管癌固定野调强临床计划（计划A），该组计划的设计条件为:无RapidPlan商业证书且不支持光子优化算法的Eclipse10.0计划系统、6 MV能量光子、Millennium 120多叶准直器、Dose Volume Optimizer优化算法和各向异性分析算法；（2）将上述病例的勾画结构传输至北大肿瘤医院的Eclipse 15.6版计划系统，保持A组计划布野、能量、多叶准直器等不变，利用RapidPlan模型在光子优化算法和光子剂量算法条件下，生成正常优化环境下的自动计划评估参考基准（计划C）；（3）作为无商业证书环境下的RapidPlan远程应用尝试，将北大肿瘤医院RapidPlan模型预测的优化条件（不包含计划结果）传回至绵阳三院计划系统，利用低版本Dose Volume Optimizer优化算法和各向异性分析算法，在不进行人工干预的情况下获得（计划B）；（4）比较A、B、C计划之间的剂量学差异。结果:无商业证书环境下跨中心应用自动计划模型具有可行性，3组计划均符合临床要求，自动计划在设计效率和一致性方面具备优势。结论:RapidPlan可以在无商业证书的低版本环境下实现跨治疗技术远程自动计划设计，提高基层医院的临床工作效率，缩小地域间医疗服务水平的差异。

Implementation of cross-institutional automated planning for esophageal cancer in the absence of commercial license

Abstract: Objective RapidPlan improves efficiency, quality and consistency of treatment planning. However, its application in local hospitals is limited by resources such as version and commercial license of Eclipse system. Taking esophageal cancer as an example, the feasibility and dosimetric performance of cross-intuitional automated planning using RapidPlan based on different treatment techniques are investigated in cases of no commercial license and planning system of lower version, aiming to benefit more patients without extra cost for hospitals. Methods (1) Twenty-six clinical IMRT plans for esophageal cancer of various sections were selected from the Third Hospital of Mianyang (Plan A). These plans were created using Eclipse V10.0 system without RapidPlan commercial license and no support for photon optimization algorithm. All plans were optimized using 6 MV photon, Millennium 120 MLC, dose volume optimizer and anisotropic analytical algorithm. (2) The aforementioned cases were transferred to Peking University Cancer Hospital where RapidPlan model was used to re-optimize these plans without changing the beam, energy and MLC settings. Photon optimization and photon dose calculation algorithms were used for the re-optimization. The automated plans obtained in the normal optimization environment of RapidPlan were taken as the reference (Plan C). (3) As an attempt to apply RapidPlan remotely in the absence of commercial license, the optimization parameters (without including optimization results) predicted by RapidPlan model were transferred to the treatment planning system in the Third Hospital of Mianyang. The plans were reoptimized using dose volume optimizer and anisotropic analytical algorithm of Eclipse system of lower version, without human intervention (Plan B). (4) Dosimetric parameters were compared among plans A, B and C. Results The cross-institutional automated treatment planning in the absence of commercial license was technically feasible. All plans were clinically acceptable, but automated plan performed better in terms of efficiency and consistency. Conclusion Without commercial license, RapidPlan can be applied remotely using treatment planning system of lower version and different treatment techniques, which enhances clinical efficiency of local hospitals, and reduces the inter-regional variabilities of medical service.