不同配准算法对晚期食管癌自适应放疗累积剂量的影响

Effects of different registration algorithms on cumulative dose of adaptive radiotherapy for advanced esophageal cancer

目的:研究不同配准方法对晚期食管癌自适应放疗累积剂量的影响并为临床应用提供指导。方法:选取11例不可手术的晚期食管癌患者,将两程放疗计划均导入Velocity软件,并以第二程计划CT图像(CTB)作为primary image,第一程计划CT图像(CTA)作为secondary image,用4种配准方法,即Rigid(Rd)、Deformable Multi Pass(DMP)、Rigid+Deformable Multi Pass(Rd+DMP)、Rigid+Extended Deformable Multi Pass(Rd+EDMP)进行图像配准及剂量叠加。采用危及器官(OAR)相似性指数(DSC)来评价不同配准方式的精确度,并采用配对t检验和秩和检验分析不同配准方式下靶区和OAR累积剂量的差异。结果:DMP和Rd+DMP配准下肝、脊髓和脊髓外扩3mm的DSC均高于Rd+EDMP、Rd(P<0.05),Rd方法下心脏的DSC略大于Rd+EDMP(P=0.02)。不同配准方法下靶区累积剂量表现为:Rd+DMP、Rd+EDMP方法下PGTV的D50%、Dmean略低于Rd方法(P=0.035,0.044)。Rd+DMP方法下PGTV的V100%稍高于DMP(P=0.026),且比Rd+EDMP方法高0.653%(P=0.008)。Rd+DMP方法下PGTV-nd的D98%、V105%略高于DMP(P=0.046,0.023),Rd方法下PGTV-nd的V100%分别比DMP、Rd+DMP方法高2.118%、2.137%(P=0.028,0.037)。PTV1在不同配准方法下累积剂量均无统计学意义(P>0.05)。4种不同配准方法下OAR累积剂量表现相对一致,DMP方法下肺的V10 Gy、V20 Gy、V30 Gy、V40 Gy、Dmean及心脏的V20 Gy、V30 Gy、V40 Gy、Dmean均低于Rd方法(P<0.05)。同样,DMP方法所得脊髓的Dmean均分别比Rd+DMP、Rd+EDMP方法低(P<0.05)。肝和脊髓外扩3 mm在以上4种配准方法下累积剂量的剂量学分析中,各参数均无统计学意义。结论:DMP、Rd+DMP配准精度相对较优、且DMP方法下OAR累积剂量一致性低于其它3种配准方法,故DMP方法可作为晚期食管癌自适应放疗累积剂量评估的优选方案。

Objective To study the effects of different registration algorithms on the cumulative dose of adaptive radiation therapy for advanced esophageal cancer and to provide guidance for clinical application.Methods Eleven patients with inoperable advanced esophageal cancer were selected and their two-course radiotherapy plans were imported in Velocity sofware,with CTA image obtained in the second course of treatment as primary image and CTB obtained in the first course of treatment as secondary image.Four registration algorithms,namely Rigid(Rd),Deformable Multi Pass(DMP),Rigid+Deformable Multi Pass(Rd+DMP),Rigid+Extended Deformable Multi Pass(Rd+EDMP),were chosen for image registration and dose accumulation.The registration accuracies of different registration algorithms were evaluated by the Dice similarity index(DSC)of organs-at-risk(OAR),and the differences in the cumulative dose of target areas and OAR under the registrations by different algorithms were analyzed by paired-t test and rank sum test.Results The DSC of liver,spinal cord and spinal cord with an expansion of 3 mm(spinal cord03)under the registration by DMP and Rd+DMP were higher than those obtained by Rd+EDMP and Rd(P<0.05),and the heart DSC obtained by Rd was slightly higher than that obtained by Rd+EDMP(P=0.02).The cumulative dose of target areas under the registrations by different algorithms showed that the D50% and Dmean of PGTV under the registrations by Rd+DMP and Rd+EDMP were slightly lower than those under the registration by Rd(P=0.035,P=0.044).The V100% of PGTV obtained by Rd+DMP was slightly higher than that obtained by DMP(P=0.026)and was 0.653% higher than that obtained by Rd+EDMP(P=0.008).The D98% and V105% of PGTV-nd obtained by Rd+DMP were slightly higher than those obtained by DMP(P=0.046,P=0.023),and the PGTV-nd V100% under the registration by Rd were 2.118% and 2.137% higher than that under the registration by DMP and Rd+DMP,respectively(P=0.028,P=0.037).There was no significant difference in the cumulative dose of PTV1 among all registrations(P>0.05).The cumulative doses of OAR under the 4 different registration algorithms were relatively consistent.The V10 Gy,V20 Gy,V30 Gy,V40 Gy,Dmean of the lungs and the V20 Gy,V30 Gy,V40 Gy and Dmean of the heart under the registration by DMP were lower than those under the registration by Rd(P<0.05).Similarly,the Dmean of the heart and spinal cord obtained by DMP was lower than that obtained by Rd+DMP and Rd+EDMP(P<0.05).Moreover,there was no statistical difference in the cumulative doses of liver and spinal cord03.Conclusion DMP and Rd+DMP have a relatively high registration accuracy,and the consistence of OAR cumulative dose under the registration by DMP was the lowest.Therefore,DMP can be used for the evaluation of cumulative dose of adaptive radiotherapy for advanced esophageal cancer.