基于三种CT图像勾画的非小细胞肺癌靶体积比较研究

A comparative study of target volumes based on three CT images in non-small cell lung cancer

目的比较基于三维CT（3DCT）、四维CT（4DCT）和锥形束CT（CBCT）图像勾画所得非小细胞肺癌（NSCLC）靶区位置和体积差异。方法31例周围型NSCLC患者，序贯完成胸部3DCT和4DCT扫描，基于3DCT制定放疗计划，放疗首次拍摄CBCT，并基于骨性标志配准校正。在3DCT、4DCT的50％时相、最大密度投影（MIP）、CBCT图像上勾画得到GTV3D、GTV4D50％、IGTVMIP和IGTVCBCT对组间位移比较行Wilcoxon秩和检验，靶区位置及包含度比较行配对t检验，肿瘤三维运动相关性行Pearson法分析。结果肺上叶组GTV3D、GTV4050%、IGTVMIP与IGTVcBcT比值分别为0．77、0．84和1．10（P=0．004、0．005、0．07）；中下叶组比值分别为0．67、0．65和1．17（P=0．001、0．001、0．020）。全组患者GTV4050%与IGTVCBCT比值与肿瘤三维运动呈负相关（P=0．012）。全组患者IGTVcBcT对GTV3D、GTV4050％、IGTVMIP包含度分别为0．65、0．65和0．62，IGTVcBcT对GTVMIP包含度与IGTVcBcT对GTV3D或GTV4050％包含度差异无统计学意义（P=0．375、0．167），而GTV3D、GTV4D50％、IGTVMIP对IGTVc哪包含度分别为0．47、0．49和0．67，IGTVMIP对IGTVc哪包含度大于GTV3D或GTV4050%。对IGTVCBCT包含度（P=0．000、0．000）。结论CBCT图像包含的肿瘤运动信息量明显大于3DCT图像，但略小于4DCT韵MIP图像。即使3DCT、4DCT与CBCT配准校正后也有可能导致较严重的脱靶，这是基于CBCT进行循证靶区和计划修正所需注意的。

Objective To determine the positional and volumetric differences between the target volumes delineated based on three-dimensional computed tomography （3DCT） , four-dimensional computed tomography （4DCT）, and cone-beam computed tomography （CBCT） in non-small cell lung cancer （NSCLC）. Methods Thirty-one patients with peripheral NSCLC sequentially underwent 3DCT and 4DCT simulation scans of the thorax during normal breathing. A 3D conformal treatment plan was created based on 3DCT. Before the first treatment, CBCT was performed and registered to the planning CT using bony anatomy registration. All contours were delineated by a radiation oncologist using the same contouring protocol. GTV3D and GTV4050% were contoured based on 3DCT and end-expiration phase （50% phase） of 4DCT, respectively;internal GTVs （IGTVMIv and IGTVcBcT ） were contoured based on maximum intensity projection （MIP） of 4DCT and CBCT, respectively. The differences in the position, size, and degree of inclusion （DI） between these volumes were determined by Vilcoxon rank-sum test and paired t-test. The Pearson test was used for the correlation analysis on 3D motion vector of GTV. Results The mean size ratio of GTVaD, GTV4050% , and IGTVMIP to IGTVCBCT were 0. 77, 0. 84, and 1.10 （z = - 2.91, P = 0. 004 ;z = - 2. 79, P = 0. 005 ;z = - 1.81, P = 0. 070） for tumors in the upper lobe of the lung and 0. 67, 0. 65, and 1. 17 （ z = - 3.30, P = 0.001 ; z = - 3.30, P = 0.001 ; z = - 2. 32, P = 0. 020 ） for tumors inthe middle/lower lobe of the lung. GTV4050%/IGTVCBCT ratio was negatively correlated with the 3D motion vector of GTV （ r = -0. 45, P =0. 012） in all the patients. The mean Dis of GTV3D, GTV4050% , and IGTVMIP in IGTVcBcT were 0. 65, 0. 65, and 0. 62, respectively, and the DI of GTV3D or GTV4050% in IGTVCBCT showed no significant difference from that of IGTVMIP in IGTVCBCT （ t = - 0. 90, P = 0. 375 ; t = - 1.42, P = 0. 167 ） ; the mean Dis of IGTVCBCT in GTV3D, GTV4050% , and IGTVMIP were 0.47, 0. 49, and 0. 67, respectively, and the DI of IGTVCBCT in IGTVMIP was significantly greater than that of IGTVCBCT in GTVaD or GTV4050% （ t = -8.28, P =0. 000;t = -5. 70, P =0. 000）. Conclusions CBCT can help to acquire significantly more information on tumor motion than 3DCT and end-expiration phase of 4DCT, but with slightly less information than 4DCT MIP. The use of 3DCT or 4DCT registered to CBCT based on bony anatomy may still result in a serious target miss in radiotherapy, which should be focused on when we perform adaptive radiotherapy and rectify treatment planning based on CBCT.