颈椎形变配准与剂量叠加MIM软件临床应用可靠性研究  

Reliability of MIM software in the study of cervical vertebra deformation registration and dose superposition

作　　者：石慧烽[1] 涂文勇[1] 孔月虹 胡海生[1] 冯琨[1] 徐璇丽 SHI Hui- feng;TU Wen-yong;KONG Yue-hong;HU Hai-sheng;FENG Kun;XU Xuan-li(Department of Oral and Maxillo ficial-Head and Neck Oncology, Ninth People ~ s Hopital of Shanghai Jiaotong University School of Medicine ,Shanghai 200011 ,P. R. China)

机构地区：[1]上海交通大学医学院附属第九人民医院口腔颌面-头颈肿瘤科,上海200011

出　　处：《中华肿瘤防治杂志》2018年第5期365-369,共5页Chinese Journal of Cancer Prevention and Treatment

基　　金：上海市科委医学引导类项目(14411967900)

摘　　要：目的形变配准在靶区勾画领域中应用广泛,配准软件层出不穷。本研究通过分析Medical image merge(MIM)软件在颈椎形变配准以及剂量叠加的可靠性,确定其在放疗的价值。方法选取上海交通大学医学院附属第九人民医院口腔颌面头颈肿瘤科在2015-05-01-2017-05-31放疗期间复发的舌癌患者,分别运用MIM软件完成刚性配准与形变配准,进行相似度验证和剂量叠加验证。相似度验证:获取每位患者的首次定位CT以及复发时定位CT,将两组CT图像输入MIM软件,每个患者都做刚性配准和形变配准,利用DICE流程分别计算出刚性配准和形变配准的Dice系数、Jaccard系数和Jausdorff距离,比较两种配准结果。放疗剂量叠加验证:将两组CT图像以及颈椎勾画分别输入到Pinnacle的治疗计划系统中进行计算,得到每个椎体的平均照射剂量,并将两组椎体平均剂量相加得到每个椎体叠加照射剂量,再将两组CT图像、椎体勾画以及剂量文件传回MIM软件中,在MIM上分别运行刚性剂量叠加以及形变剂量叠加流程得到得到叠加剂量,比较放射治疗计划系统(radiotherapy treatment planning system,TPS)计算结果跟MIM软件得到的结果差异。结果相似度验证中,刚性配准中Dice系数=0.73±0.09,Jaccard系数=0.58±0.14,Hausdorff距离=0.68±0.21;在形变配准中,Dice系数=0.85±0.03,Jaccard系数=0.75±0.04,Hausdorff距离=0.60±0.13。两组数据差异有统计学意义,P<0.05。剂量叠加验证中,刚性配准得到的叠加剂量跟TPS计算的剂量平均差别为1.38%,但是剂量差别>3%有16%,最大剂量差别>9%;而形变配准得到的叠加剂量跟TPS计算得到的剂量平均差距为0.05%,>3%为0,<1%为88.1%。结论相对于刚性配准,MIM软件的形变配准功能可以将CT图像进行更好的融合,其剂量误差<3%允许范围,因此,其剂量叠加功能具有实际应用价值。OBJECTIVE Deformation registration is widely used in target area depiction.There are so many software of registration.This study aims to observe the Medical image merge（MIM）software in cervical deformation registration and the reliability of dose superimposed to determine its value in radiotherapy.METHODS Twenty recurrence of tongue cancer patients underwent radiotherapy were chosen from 2015-05-01 to 017-05-31,in Department of Oral and Maxilloficial-Head and Neck Oncology,Ninth People＇s Hopital of Shanghai Jiaotong University School of Medicine.The patients were randomly divided into two groups：routine registration and deformation registration using MIM software.The similarity verification and dose superposition verification were performed.Similarity verification：Two sets of CT images that were taken before and after recurrence were input into MIM software,and the rigidity and deformation were registered.The Dice coefficient,Jaccard coefficient,Hausdorff distance,and 2 kinds of rigid registration and deformation registration were calculated by DICE process.Radiotherapy dose superposition verification：Two sets of CT images and cervical vertebrae were input to Pinnacle＇s treatment planning system to calculate the average dose of each vertebral body,and the average dose of the two groups were added and then the two groups of CT images,vertebral outline and doses were processed back to the MIM software,respectively.In the MIM,the rigid dose superposition and deformation dose superposition were processed to get the superposition of the dose.Compared with the TPS（Radiotherapy Treatment Planning System）,calculation results with the MIM,the results of the software were different.RESULTS In the similarity verification,the Dice coefficient in the rigid registration was 0.73±0.09,the Jaccard coefficient was 0.58±0.14,the Hausdorff distance was 0.68±0.21.In the deformation registration,the Dice coefficient was 0.85±0.03,the Jaccard coefficient was0.75±0.04,Hausdorff distance was 0.60±0.13,and there w

关 键 词：MIM软件 形变配准 刚性配准 剂量叠加 Dice系数 Jaccard系数 HAUSDORFF距离 

分 类 号：R730.55[医药卫生—肿瘤]