两种不同优化算法在胸部肿瘤调强放射治疗计划中的对比研究

Study on Two Optimization Algorithms for Radiotherapy Treatment Planning of Thoracic Tumor

目的:研究直接子野优化(DMPO)算法与多目标优化(MCO)算法在胸部肿瘤放射治疗计划设计优化过程中的差异。方法:选取20例食管癌患者,将患者在Pinnacle3计划系统的CT图像、勾画的靶区、设计参数和剂量信息等资料,通过DicomRT协议从Pinnacle3计划系统传输到Ray Station计划系统中。保持照射野设置、处方剂量不变,改用MCO算法重新优化治疗计划,比较两种优化算法在剂量分布、靶区适形指数(CI)与靶区均匀指数(HI)、危及器官(OAR)剂量、计划设计与执行效率的差异性。结果:两种算法优化的计划结果均能基本满足临床剂量要求,与DMPO相比,MCO计划靶区PGTV适形指数CIpgtv和均匀指数HIpgtv以及靶区PTV的适形指数CIptv均差别不大,但危及器官的受量明显小于前者,其中肺的V10、V20、V30、平均剂量Dmean-l,心脏的V30、V40和平均剂量Dmean-h以及脊髓最大剂量D1cm3-s和平均剂量Dmean-s,均存在显著性差异;从计划的设计时间上来看,MCO计划组明显短于DMPO计划组,而从执行效率上来看,两者差异不大。结论:对于食管癌同期加量三维调强放疗计划,与DMPO算法相比,MCO算法可得到更低的OAR剂量,并且在临床应用上明显提高工作效率。

Objective To study the different performance of the two optimization algorithms, DMPO and MCO, in IMRT plan- ning of thoracic tumor. Methods The 20 cases with esophageal carcinoma was randomly chosen in the trail. The IMRT plan was done on Pinnacle3 system with the DMPO optimization algorithm. Then the plan was transferred into the RayStation system via DicomRT protocol. All physics parameter and the prescription dose was kept unchanged, the treatment plan was reopti- mized with the MCO algorithm on RayStation system. The performance of two different algorithms has been fully investigated on the quantification such as target dose distribution, conformity index（CI）, homogeneity index（HI）, organ at risk（OAR） dose, time consumption and work efficiency. Results All planning results with two different optimization algorithms can basically meet the clinical requirement. There is no significant difference observed in the PGTV dose distribution, HI of the PGTV, CI of the PTV and PGTV between the DMPO and MCO optimization algorithm. But the less dose of OAR, such as V10, V20, V30 of the lung, mean dose of the heart and spinal cord, was shown in the IMRT plan with MCO optimization algorithm. The consum- ption of planning time is obviously less with MCO optimization algorithm. Conclusion All the IMRT plan with two optimiza- tion algorithms can meet the clinical requirement, but less OAR dose can get with the MCO algorithm.