摘要目的：应用MRI T1WI、T2WI、T1WI增强信号强度直方图分析，对弥漫大B淋巴瘤与胶质母细胞瘤进行肿瘤全域分析并评估其对二者鉴别诊断的价值。材料与方法：回顾性分析经手术病理证实的16例弥漫大B淋巴瘤（Large B cell lymphoma，LYM）（男9例，女7例，平均年龄（57.9±10.1）岁）和20例胶质母细胞瘤（Glioblastoma，GBM）（男10例，女10例，平均年龄（57.8±11.8）岁）患者的术前MRI资料，在包含肿瘤实质的每一层T1WI、T2WI、T1WI增强信号强度图上勾画感兴趣区（Region of interest，ROI），得到3D ROI的T1WI、T2WI、T1WI增强信号强度直方图信息及参数，包括最小值、最大值、平均值、标准偏差、方差、偏度、峰度、均匀性、熵值、第10百分位数、第25百分位数、第50百分位数、第75百分位数、第90百分位数等，进行组间比较，利用受试者操作特性曲线（Receiver operating characteristic curve，ROC）来确定直方图参数对于二者的诊断能力。应用Logistic回归分析模型得到联合变量（Joint variable，JV），利用ROC曲线来确定其诊断能力。结果：T1WI增强信号强度直方图参数中LYM最小值、均匀性高于GBM，LYM峰度值低于GBM。LYM偏度值为负值，与其不同的是GBM偏度值为正值，且均具有显著差异性（P<0.05）。T1WI、T2WI信号强度直方图参数中，LYM标准偏差、方差均低于GBM，LYM均匀性高于GBM，且均具有显著差异性（P<0.05）。T1WI增强信号中以偏度0.007 8为阈值的曲线下面积（Area under the curve，AUC）最大（AUC=0.819），诊断能力最佳，敏感性及特异性分别为65%、87.5%。T1WI信号中以标准偏差78.02、方差6 087.64为阈值的AUC（0.787、0.787）最大，敏感性及特异性分别为75%、81.2%。T2WI信号中以标准偏差452.68、方差204 937为阈值的AUC（0.7、0.7）最大，敏感性及特异性分别为55%、81.2%。JV以0.55为阈值的AUC（0.881）最大，敏感性及特异性分别为80%、87.5%。结论：基于肿瘤全域ROI T1WI、T2WI以及T1WI增强信号直方图可以为LYM与GBM的鉴别提供更多信息，联合应用直方图参数可提高对二者的诊断能力，为鉴别这两种病变提供可靠的客观依据。
Abstract：Objective: To elucidate differences between large B cell lymphoma(LYM) and glioblastoma(GBM) using histogram analysis of T1WI, T2WI and contrast enhanced T1WI signal intensity maps based on entire tumor region, and then to evaluate the differential diagnosis value between them. Materials and Methods: Sixteen patients with LYM(9 males, 7 females; mean age 57.9±10.1 years old) and twenty patients with GBM(10 males, 10 females; mean age 57.8±11.8 years old) were enrolled in this retrospective study, and all tumors were pathologically confirmed. All patients undertook conventional MRI scan including T1WI, T2WI and contrast enhanced T1WI. Regions of interest(ROIs) containing the entire tumor were manually drawn in each slice of the T1WI, T2WI and contrast enhanced T1WI signal intensity maps. The histogram and all its parameters were obtained from the 3D ROI of T1WI, T2WI and contrast enhanced T1WI signal intensity using Omni-Kinetics software. Histogram parameters included min intensity, max intensity, mean value, standard deviation, variance, skewness, kurtosis, uniformity, entropy, the 10th, 25th, 50th, 75th and 90th percentiles. All parameters were compared between groups. The joint variable(JV) was calculated using the logistic regression analysis model. Receiver operating characteristic curve(ROC) was used to assess the differential ability of all parameters between LYM and GBM. Results: In the contrast enhanced T1WI signal histogram parameters, min intensity and uniformity of LYM were higher than GBM, but kurtosis of LYM were lower than GBM(P＜0.05). The negative skewness was present in LYM and that was positive in GBM(P＜0.05). In T1WI and T2WI signal intensity histogram parameters, standard deviation and variance of LYM were lesser than those of GBM, but uniformity was increased in LYM than GBM(P<0.05). According to ROC analysis, the skewness(cutoff value=0.007 8, area under the curve(AUC)=0.819) was considered as the best parameter for diagnosis of LYM and GBM in the contrast enhanced T1WI, with the sensitivity of 65% and specificity of 87.5%. Moreover, in the T1WI histogram parameters, the standard deviation and variance were the most optimal parameters(cutoff value=78.02, 6 087.64; AUC=0.787, 0.787) for distinguishing LYM and GBM with sensitivity of 75% and specificity of 81.2%. In the T2WI histogram parameter, the standard deviation and variance were the most optimal parameters(cutoff value=452.68, 204 937; AUC=0.7, 0.7) for distinguishing LYM and GBM with sensitivity of 55% and specificity of 81.2%. And when optimal cut point of the joint variable was 0.55 for diagnosis of LYM and GBM, the area under the AUC(0.881) was maximum with the sensitivity of 80% and specificity of 87.5%. Conclusion: Histogram analysis based on the ROI of tumor of T1WI, T2WI and contrast enhanced T1WI signal can provide more information to identify large B cell lymphoma and glioblastoma, and can improve the diagnostic ability of them combined with histogram parameters. It can provide a reliable objective basis for the identification of them.
韩 亮1，苗延巍1，董俊伊1，李晓欣1，刘杨颖秋1，田诗云1，王微微1，郭 妍2，宋清伟1，刘爱连1. 基于肿瘤全域的常规MRI参数直方图分析对弥漫大B淋巴瘤与胶质母细胞瘤的鉴别[J]. 中国临床医学影像杂志, 2018, 29(12): 837-843.
HAN Liang1, MIAO Yan-wei1, DONG Jun-yi1, LI Xiao-xin1, LIU Yang-yingqiu1, TIAN Shi-yun1, WANG Wei-wei1, GUO Yan2, SONG Qing-wei1, LIU Ai-lian1. Histogram analysis of conventional MRI parameters for differentiating glioblastoma from large B cell lymphoma based on whole tumor measurement . JOURNAL OF CHINA MEDICAL IMAGING, 2018, 29(12): 837-843.
[1]Tozer DJ, Jager HR, Danchaivijitr N, et al. Apparent diffusion coeffcient histograms may predict low-grade glioma subtype[J]. NMR Biomed, 2007, 20(1): 49-57.
[2]Murayama K, Nishiyama Y, Hirose Y, et al. Differentiating between central nervous system lymphoma and high-grade glioma using dynamic susceptibility contrast and dynamic contrast-enhanced MR imaging with histogram analysis[J]. Magn Reson Med Sci, 2018, 17(1): 42-49.
[3]Anvari A, Halpern EF, Samir AE. Statistics 101 for radiologists[J]. Radiographics, 2015, 35(6): 1789-1801.
[4]Bartlett JW, Frost C. Reliability, repeatability and reproducibility: analysis of measurement errors in continuous variables[J]. Ultrasound Obstet Gynecol, 2008, 31(4): 466-475.
[5]苗炜宇，张岚. 磁共振弥散加权成像在鉴别脑内淋巴瘤与高级别胶质瘤中的应用价值[J]. 中国实用神经疾病杂志，2016，19（8）：65-67.
[6]季学满，卢光明，张宗军，等. 原发性脑淋巴瘤与高级别脑胶质瘤的MR灌注成像对照研究[J]. 临床放射学杂志，2008，27（9）：1155-1158.
[7]卢昊，冯全志，程乾胜，等. DSC-MRI鉴别诊断胶质母细胞瘤、单发脑转移瘤及脑淋巴瘤[J]. 中国医学影像技术，2017，33（8）：1185-1189.
[8]Halshtok Neiman O, Sadetzki S, Chetrit A, et al. Perfusionweighted imaging of peritumoral edema can aid in the differential diagnosis of glioblastoma mulltiforme versus brain metastasis[J]. Isr Med Assoc, 2013, 15(2): 103-105.
[9]Choi YS, Ahn SS, Kim DW, et al. Incremental prognostic value of ADC histogram analysis over MGMT promoter methylation status in patients with glioblastoma[J]. Radiology, 2016, 281(1): 175-184.
[10]张胜，李玉林，黄送，等. 增强T1WI直方图在胶质母细胞瘤和脑单发转移瘤鉴别诊断中的应用[J]. 中国医学影像学杂志，2017，25（2）：89-92.
[11]Xu XQ, Hu H, Su GY, et al. Utility of histogram analysis of ADC maps for differentiating orbital tumors[J]. Diagn Interv Radiol, 2016, 22(2): 161-167.
[12]Suo ST, Chen XX, Fan Y, et al. Histogram analysis of apparent diffusion coefficient at 3.0T inurinary bladder lesions: correlation with pathologic fndings[J]. Acad Radiol, 2014, 21(8): 1027-1034.
[13]Woo S, Cho JY, Kim SY, et al. Histogram analysis of apparent diffusion coefcient map of diffusion-weighted MRI in endometrial cancer: a preliminary correlation study with histological grade[J]. Acta Radiol, 2014, 55(10): 1270-1277.
[14]Ahn SJ, Choi SH, Kim YJ, et al. Histogram analysis of apparent diffusion coefcient map of standard and high B-value diffusion MR imaging in head and neck squamous cell carcinoma: a correlation study with histological grade[J]. Acad Radiol, 2012, 19(10): 1233-1240.
[15]Ryu YJ, Choi SH, Park SJ, et al. Glioma: application of whole-tumor texture analysis of diffusion-weighted imaging for the evaluation of tumor heterogeneity[J]. PLoS One, 2014, 9(9): e108335.
[16]Zhang YD, Wang Q, Wu CJ, et al. The histogram analysis of diffusion-weighted intravoxel incoherent motion(IVIM) imaging for differentiating the gleason grade of prostate cancer[J]. Eur Radiol, 2015, 25(4): 994-1004.
[17]Kunimatsu A, Kunimatsu N, Kamiya K, et al. Comparison between glioblastoma and primary central nervous system lymphoma using MR image-based texture analysis[J]. Magn Reson Med Sci, 2018, 17(1): 50-57.
[18]Baek HJ, Kim HS, Kim N, et al. Percent change of perfusion skewness and kurtosis: a potential imaging biomarker for early treatment response in patients with newly diagnosed glioblastomas[J]. Radiology, 2012, 264(3): 834-843.
[19]董旭，李云华，明照亭，等. 多发性脑胶质瘤与原发性中枢神经系统淋巴瘤的MRI鉴别诊断[J]. 齐鲁医学杂志，2015，30（1）：15-18.
[20]Meyer HJ, Schob S, Münch B, et al. Histogram analysis of T1-weighted, T2-weighted, and postcontrast T1-weighted images in primary CNS lymphoma: correlations with histopathological findings-a preliminary study[J]. Mol Imaging Biol, 2018, 20(2): 318-323.
[21]Martin B, Paesmans M, Mascaux C, et al. Ki-67 expression and patients survival in lung cancer: systematic review of the literature with meta-analysis[J]. Br J Cancer, 2004, 91(12): 2018-2025.
[22]He X, Chen Z, Fu T, et al. Ki-67 is a valuable prognostic predictor of lymphoma but its utility varies in lymphoma subtypes: evidence from a systematic meta-analysis[J]. BMC Cancer, 2014, 14: 153.
[23]颜虹. 医学统计学[M]. 北京：人民卫生出版社，2010：29-32.
[24]黄斌，黄永杰. 原发性脑淋巴瘤的影像学表现与病理学对照研究[J]. 医学影像学杂志，2008，18（11）：1217-1220.
[25]孙振国，汪秀玲，朱辉，等. 纹理分析在原发性脑淋巴瘤与高级别胶质瘤鉴别诊断中的应用价值[J]. 临床放射学杂志，2017，36（9）：1229-1234.
[26]范亦龙，陈绪珠，戴建平. 脑胶质瘤动态变化的形态学观察及其临床意义[J]. 医学影像学杂志，2010，20（4）：453-456.
[27]Price SJ, Jena R, Burnet NG, et al. Improved delineation of glioma margins and regions of infltration with the use of diffusion tensorn imaging: an image-guided biopsy study[J]. AJNR, 2006, 27(9): 1969-1974.
[28]刘晓玉，肖道雄，何艳枚，等. 联合应用DWI及PWI对脑内肿瘤及瘤周水肿的比较研究[J]. 临床放射学杂志，2017，36（7）：928-933.
[29]Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Oganization classifcation of tumors of the central nervous system: a summary[J]. Acta Neuropathol, 2016, 131(6): 803-820.
