直肠癌术前同步推量RapidPlan模型用于术后单一处方计划优化的改造测试

Feasibility study of applying a RapidPlan model trained with pre-surgical rectal cancer SIB plans to the optimization of post-surgical non-SIB plans

目的:测试直肠癌术前同步推量RapidPlan模型用于术后单一处方计划优化的可行性,探索改造和拓展已有模型的应用方法。方法:(1)对现有直肠癌术前同步推量41.8 Gy(PTV)/50.6 Gy(PGTV)RapidPlan模型进行改造;(2)复制直肠癌术后单一PTV处方临床计划18例(45 Gy的7例,50 Gy的11例),保持布野条件、能量、加速器配置及原有剂量算法等不变,将PTV分别匹配模型中的PGTV和PTV,预测可实现的DVH区间并重新优化;(3)将自动计划靶区剂量归一至与临床计划相似后比较各剂量学参数。结果:单一处方靶区匹配模型中的PTV会导致严重的剂量热区,匹配PGTV的RapidPlan计划质量和临床计划相似或略好,但其危及器官剂量学参数的标准差均小于临床计划。结论:直肠癌术前同步推量Rapid Plan模型可用于术后单一处方放疗计划的自动优化并且其质量一致性更好,但需将单一靶区匹配给模型中最高处方剂量对应的靶区结构。

Objective To investigate the feasibility of applying a RapidPlan model trained with pre-surgical rectal cancer simultaneous integrated boost(SIB)plans to the optimization of post-surgical non-SIB plans,and to explore the methods to improve the existing models and extend their applications.Methods An existing RapidPlan model for pre-surgical rectal cancer SIB plans(PTV/PGTV:41.8 Gy/50.6 Gy)was modified and used to re-optimize 18 clinical post-surgical non-SIB plans among which 7 and 11 cases were prescribed with 45 Gy and 50 Gy to PTV,respectively,without changing the original beam geometries,energies,accelerator configuration and dose volume calculation algorithm.The target structures were matched with PTV and PGTV in the model.The realizable DVH intervals were predicted,and the re-optimized plans were renormalized to achieve similar target coverage of the clinical plans before dosimetric comparison.Results Severe hot spots were caused by matching target areas of non-SIB plans to PTV in the model,while re-optimized plans could achieve similar or better plan quality with smaller standard errors of organs-at-risk dosimetric parameters,compared with the clinical plans.Conclusion RapidPlan model for pre-surgical rectal cancer SIB plans can be applied to the automatic optimization of post-surgical non-SIB plans with higher consistency in plan quality,but the target structure is needed to be matched to the high prescription target structure in SIB model.