基于3D-CT与4D-CT勾画保留乳房手术后全乳靶区的比较研究

Comparison study of clinical target volumes of whole breast after breast-conserving surgery based on three-dimensional CT and four-dimensional CT images

目的探讨基于三维CT(3D-CT)与四维CT(4D-CT)勾画的乳腺癌保留乳房手术后全乳靶区(CTV)的差异性。方法对13例保留乳房手术后患者于CT模拟定位时序贯完成胸部3D-CT和4D-CT扫描,并依据实时位置管理系统(RPM)同步采集的呼吸信号将每个呼吸周期的4D-CT图像分为10个呼吸时相。将图像传入Eclipse计划系统,以4D-CT的吸气末(T0)时相为基准,其余9个时相的9套图像(T10、T20、30……T90)、最大密度投影图像(MIP)及3D-CT图像分别与之配准。同一勾画者分别于两个不同时间,在4D-CT的T0图像上勾画源于3D-CT、T0、呼气中(T20)、呼气末(T50)及MIP图像上的全乳靶区。之后,在4D-CT的T0图像上勾画源于3D-CT、4D-CT及MIP图像上的全乳靶区,并分别定义为CTV3D、CTV0、CTV10……CTV90和CTVMIP。最后,将4D-CT的CTV0、CTV10、CTV20……CTV90融合得到融合靶区(internalclinicaltargetvolume,ICTV)。比较4D-CT不同时相图勾画的全乳靶区后,选取其中具有代表性的T0、T20、T50、MIP图像与3D-CT图像相比。比较同一勾画者的勾画差异性以及基于3D-CT与4D-CT勾画的全乳CTV体积、匹配指数(MI)和包含度(DI)的差异性。计量资料比较采用t检验或Friedman、Wilcoxon秩和检验。结果无论基于3D-CT还是基于4D-CT,同一放射治疗医师勾画的靶区体积差异无统计学意义(P均>0.050)。呼吸运动对4D-CT10个时相的CTV体积大小无明显影响(P>0.050)。CTV3D、CTV0、CTV20、CTV50、CTVMIP体积的中位数分别为708.11、721.29、725.04、723.89、728.69cm3。CTV3D与CTV0、CTV20、CTV50、CTVMIP体积差异均无统计学意义(P均>0.050);CTV3D与CTV0、CTV20、CTV50的MI中位数分别为0.88、0.86和0.86,4D-CT不同时相CTV与CTV3D的MI差异无统计学意义(x2=0.462,P=0.794)。CTV3D对CTV0、CTV20、CTV50的DI中位数分别为0.94、0.93和0.92,CTV0、CTV20、CTV50对CTV3D的DI中位数分别为0.95、0.95和0.94,CTV3D与4D-CT不同时相CTV的DI差异无统计学意义(P均>0.050)。ICTV体积的中位数为793.56cm3,ICTV体积明显>CTV3D(Z=-3.180,P=0.001),CTV3D与ICTV的MI中位数为0.86。CTV3D对ICTV和ICTV对CTV3D的DI分别为0.91和0.96,两者之间差异有统计学意义(Z=-3.180,P=0.001)。ICTV体积明显>CTVMIP体积(Z=-3.180,P=0.001),两者之间DI差异有统计学意义(Z=-3.180,P=0.001),ICTV与CTVMIP的MI中位数为0.93。结论在勾画标准一致的情况下,同一勾画者所勾画的全乳靶区不受CT扫描方式的影响。3D-CT扫描所采集的呼吸运动信息有限,呼吸运动对内靶区(ITV)的构建影响显著,基于4D-CT扫描图像构建ITV更合理。

Objective To study the differences of the clinical target volume （CTV） based on three-dimensional CT （3D-CT） and four-dimensional CT （4D-CT） of the whole breast after breast-conserving surgery. Methods Thirteen patients after breast-conserving surgery underwent 3D-CT simulation scans followed by 4D-CT simulation scans of the thorax during free breathing. During 4D-CT scanning, real-time position management （RPM） system simultaneously recorded the respiratory signals. The CT images with respiratory signal data were reconstructed and sorted into 10 phase groups in a respiratory cycle. Data sets for 3D-CT and 4D-CT scans were then transferred to Eclipse treatment planning software. The 4D-CT image of the end-inhalation phase （TO） served as a background and the other nine phases （T10, T20, T30... q＇90 ）, maximum intensity projection （MIP） image and 3D-CT image were registered. The CTV were manually delineated on the registered images of the 3D-CT, TO, middle-exhalation （T20） , end-exhalation （TS0）, MIP images based on the TO of 4D-CT by a radiation oneologist at two different times. Then the CTV3D, CTV0, CTV,0 ... CTVMw were delineated and defined on the 3D-CT, T0, T10... MIP images based on the TO images of 4D-CT by the same radiation oneologist. All the CTVs （CTV0, CTVlo, CTV10... CTV50 ） delineated on the 10 phases of the 4D-CT images were fused into an internal clinical target volume （ICTV）. The T0, T20, T50, MIP images were selected from the CTVs of the 4D-CT to compare with the 3D-CT image. The differences of the targets delineated on the same images by the same radiation oncologist at different times were compared. The volumes of the CTVs, the matching index （MI） and the degree of inclusion （DI） were compared respectively. Results There was no difference in the CTV delineated by the same oncologist no matter based on 3D-CT or 4D-CT（ P〉0. 050）. The CTVs volumes of ten phases in 4D-CT were not impacted by respiratory movement（ P〉 0. 05 ）. The median volume of CTV3D, CTV0, CTV20, CTVso, CTVMtP were 708. 11 cm3, 721.29 cm3, 725. 04 cm3, 723.89 cm3 , and 728.69 cm3 , respectively. The volume demonstrated no significant difference between CTV3D and CTV0, CTV2o, CTVs0, CTVMIp（P〉0. 050）. The median MI of CTVaD and CTVo, CTV20, CTV50 were 0. 88, 0. 86 and 0. 86,respectively. the difference of the MI between CTV30 and CTV0, CTV20, CTV50 was not statistically significant （x2 = 0. 462, P = 0. 794）. The median DI of CTVo, CTVz0, CTV50 in CTV3D were 0. 94, 0. 93 and 0. 92, respectively. CTV3Din CTV0, CTV20, CTVsowere 0. 95, 0. 95 and 0. 94, respectively. There was no significant difference between the DI of CTV3D in CTV of the single phases of 4D-CT and the DI of CTV of the single phases in CTV3o（P〉0. 05）. The median ICTV was 793.56 cm3, larger than that of CTV3D（Z= -3. 180, P = 0. 001）and the median MI between CTV3D and ICTV was 0. 86. The median DI of ICTV in CTV3D and CTV3Din ICTV were 0. 91 and 0. 96; the difference was significant（Z = -3. 180, P = 0. 001 ）. ICTV was significantly higher than CTVMIP（Z = -3. 180, P = 0.001） and the DI between the two had a significant difference（Z = -3. 180, P = 0. 001 ）. The median MI of ICTV and CTVMIP was 0. 93. Conclusions The delineation of clinical target volume of the whole breast would not be influenced by scan mode when the CTV is delineated by the same oncologist under the same delineation criterion. The 3D-CT shows limited movement information ; the construction of internal target volume （ITV） is significantly impacted by respiratory movement. So ITV of whole breast target delineated on the 4D-CT images is more reasonable.