基于二阶导数的小跳数射野删减优化在射波刀头部计划中的应用

Application of the second derivative-based small monitor unit beam deletion optimization to CyberKnife planning of heads

目的研究射波刀治疗计划系统中不同的小跳数(MU)射野删减优化方法对颅脑肿瘤计划剂量计算结果的影响。方法选取2021年6月至2022年2月在本院治疗的17例脑转移瘤患者,针对每例患者使用射波刀VSI系统配备的MultiPlan计划系统设计计划作为无优化组,每例初始计划生成后需删减一些小MU射野,以优化计划的执行效率,形成经验组和优化组。经验组按照经验将30 MU以下的射野删掉,优化组将基于二阶导数方法计算的MU值以下的射野删掉,最后统计对比3组计划参数。主要评估的参数包括节点数、射野数、总MU数、预计治疗时间、计划靶区(PTV)所接受的2%、95%体积的剂量PTV D2、PTV D95和平均剂量PTV Dmean,脑组织的平均剂量Dmean-Brain、适形指数(CI)、新适形指数(nCI)、梯度指数(GI)、覆盖率、脑干和左右眼晶状体的最大剂量(D_(max)-BS、D_(max)-LL和D_(max)-RL),距离PTV 20和40 mm的剂量壳的平均剂量Shell20和Shell40。结果采用的两种优化方法均可以满足>98%PTV接受处方剂量照射。无优化组、经验组和优化组在节点数(H=7.97,P<0.05)和预计治疗时间(H=6.60,P<0.05)上差异有统计学意义,优化组预计治疗时间和节点数低于无优化组(P<0.05),其他参数各组之间差异无统计学意义(P>0.05)。PTV体积与经验组治疗时间(r=0.79,P<0.01)和射野数(r=0.78,P<0.01)呈中度正相关,与优化组治疗时间(r=0.69,P<0.01)和射野数(r=0.71,P<0.01)呈中度正相关。结论对于射波刀头部计划,基于二阶导数的小MU射野删减优化方法能在保证危及器官和靶区的剂量分布无显著差异的前提下,进一步地缩短治疗时间,且对PTV体积较大的计划优化效果更好。

Objective To investigate the effects of different small monitor unit(MU)beam deletion optimization method in the CyberKnife treatment planning system on the calculated planned dose to brain tumors.Methods A total of 17 patients with brain metastases treated in our hospital from June,2021 to February,2022 were selected for this study.A treatment plan was designed for each patient using the multiPlan system in the CyberKnife VSI system as the group without optimization.To improve the efficiency,the generated original plans should be optimized first by deleting some small MUs,forming an experience group and an optimization group for each patient.For the experience group,beams below 30 MU were deleted according to experience.For the optimization group,beams below the MU value calculated based on the second derivative method were deleted.Finally,the parameters of the two groups were statistically compared.The main evaluation parameters included the node number,the beam number,the total number of MUs,the estimated treatment duration,doses to 2%and 95%planning target volumes(PTV D2 and PTV D95),average dose to PTV(Dmean),average dose to brain tissue(Dmean-Brain),conformity index(CI),new conformity index(nCI),gradient index(GI),coverage,and the maximum doses to the brainstem and left and right lens(D_(max)-BS,D_(max)-LL,and D_(max)-RL),and the average doses to the dose shells 20 mm and 40 mm away from PTV(Shell20 and Shell40).Results The two optimization method met the requirements for the prescription dose delivery to more than 98%PTV.There were statistical differences in the node number(H=7.97,P<0.05)and estimated treatment duration(H=6.60,P<0.05)among the group without MP optimization,the experience group,and the optimization group,with the estimated treatment duration and node number of the optimization group less than those of the group without MP optimization(P<0.05).There were no statistically significant differences in other parameters among the three groups(P>0.05).The PTV was moderately positively correlated with the treatment duration(r=0.79,P<0.01)and beam number(r=0.78,P<0.01)of the experience group,and was also moderately positively correlated with the treatment duration(r=0.69,P<0.01)and beam number(r=0.71,P<0.01)of the optimization group.Conclusions For the CyberKnife planning of heads,the small MU beam deletion optimization method based on the second derivative can further shorten the treatment duration while ensuring no significant differences in the distribution of doses to organs at risk and targets.Moreover,this method is more effective in optimizing the plans for a large PTV volume.