High-dimensional automated radiation therapy treatment planning via bayesian optimization.
High-dimensional automated radiation therapy treatment planning via bayesian optimization.
复制标题
DOI：
10.1002/mp.16289
复制
发表时间：
2022-05
期刊：
影响因子：
3.8
通讯作者：
Qingying Wang;Ruoxi Wang;Jiacheng Liu;F. Jiang;H. Yue;Yi Du-;Hao-Nan Wu
中科院分区：
文献类型：
--
作者：
Qingying Wang;Ruoxi Wang;Jiacheng Liu;F. Jiang;H. Yue;Yi Du-;Hao-Nan Wu
目的：放射治疗计划可以看作是一个反复的超参数调整过程，以平衡相互冲突的临床目标。在这项工作中，我们研究了现代贝叶斯优化（BO）方法在高维环境下自动治疗计划问题上的性能。方法回顾性选择20例局部晚期直肠癌调强放疗（IMRT）患者作为试验病例。可调整的规划参数包括剂量目标及其相应的权重。我们实现了一个自动化的治疗计划框架，并测试了两种BO方法在治疗计划任务上的性能：一种是标准BO方法（GPEI），一种是专门针对高维问题的BO方法（SAAS-BO）。另一种无导数方法（Nelder-Mead单纯形搜索）和随机调谐方法也作为基准。将4种自动化方法的计划质量和计划效率与临床计划在靶区覆盖率和危重器官（OAR）保留方面进行比较。比较了两种BO方法的预测模型，分析了两种BO方法的不同搜索模式。结果SAAS-BO方案在靶结构的热点控制性（p=0.43$ p=0.43$）和均匀性（p=0.96$ p=0.96$）方面均优于GPEI和Nelder-Mead方案（p < 0.05$ p < 0.05$）。SAAS-BO方案和GPEI方案在符合性和剂量溢出方面均显著优于临床方案（p < 0.05$ p < 0.05$）。与临床方案相比，四种自动化方法生成的治疗方案均降低了股骨头和膀胱的评估剂量学指标。与BO方案相比，Nelder-Mead方案获得了相似的方案质量分数，但在目标热点和剂量溢出方面表现出较差的控制。对基础预测模型的分析表明，两种BO方法都能识别出相似的敏感规划参数。这项工作实现了一个基于bo的超参数调整框架，用于自动治疗计划。经过测试的两种BO方法都能够产生高质量的治疗计划，并减少治疗计划人员的工作量。模型分析也证实了所测试的治疗计划问题的内在低维性。这篇文章受版权保护。版权所有。
PURPOSE Radiation therapy treatment planning can be viewed as an iterative hyperparameter tuning process to balance conflicting clinical goals. In this work, we investigated the performance of modern Bayesian Optimization (BO) methods on automated treatment planning problems in high-dimensional settings. METHODS 20 locally advanced rectal cancer patients treated with intensity-modulated radiation therapy (IMRT) were retrospectively selected as test cases. The adjustable planning parameters included both dose objectives and their corresponding weights. We implemented an automated treatment planning framework and tested the performance of two BO methods on the treatment planning task: one standard BO method (GPEI) and one BO method dedicated to high-dimensional problems (SAAS-BO). Another derivative-free method (Nelder-Mead simplex search) and the random tuning method were also included as baselines. The four automated methods' plan quality and planning efficiency were compared with the clinical plans regarding target coverage and organs at risk (OAR) sparing. The predictive models in both BO methods were compared to analyze the different search patterns of the two BO methods. RESULTS For the target structures, the SAAS-BO plans achieved comparable hot spot control ( p = 0.43 $p=0.43$ ) and homogeneity ( p = 0.96 $p=0.96$ ) with the clinical plans, significantly better than the GPEI and Nelder-Mead plans ( p < 0.05 $p < 0.05$ ). Both SAAS-BO and GPEI plans significantly outperformed the clinical plans in conformity and dose spillage ( p < 0.05 $p < 0.05$ ). Compared with the clinical plans, the treatment plans generated by the four automated methods all made reductions in evaluated dosimetric indices for the femoral head and the bladder. The Nelder-Mead plans achieved similar plan quality scores compared with the BO plans, but exhibited poorer control in the target hot spot and dose spillage. The analysis of the underlying predictive models has shown that both BO methods have identified similar sensitive planning parameters. CONCLUSIONS This work implemented a BO-based hyperparameter tuning framework for automated treatment planning. Both tested BO methods were able to produce high-quality treatment plans and reduce the workload of treatment planners. The model analysis also confirmed the intrinsic low dimensionality of the tested treatment planning problems. This article is protected by copyright. All rights reserved.