肿瘤全域表观扩散系数诺模图诊断高级别子宫内膜癌的价值  

Value of whole tumor volume ADC nomogram in diagnosing high-grade endometrial carcinoma

作　　者：邓颖 戴强 王茵 李志豪 赵婷婷[2] 梁轶 闫斌 DENG Ying;DAI Qiang;WANG Yin;LI Zhihao;ZHAO Tingting;LIANG Yi;YAN Bin(Department of Radiology,Shaanxi Provincial Cancer Hospital,Shaanxi Xi'an 710061,China;Department of Medical Imaging,the First Affiliated Hospital of Xi'an Jiaotong University,Shaanxi Xi'an 710061,China)

机构地区：[1]陕西省肿瘤医院放射科,陕西西安710061 [2]西安交通大学第一附属医院影像科,陕西西安710061

出　　处：《现代肿瘤医学》2023年第18期3473-3480,共8页Journal of Modern Oncology

基　　金：陕西省西安市创新能力强基计划-医学研究项目(编号:21YXYJ0102)。

摘　　要：目的:探讨并验证肿瘤全域表观扩散系数(apparent diffusion coefficient,ADC)诺模图术前预测高级别子宫内膜癌(endometrial carcinoma,EC)的应用价值。方法:回顾性收集经术后病理证实为EC患者142例,按7∶3的比例分为训练组(n=99)与验证组(n=43)。所有患者术前均接受3.0-T MR检查。采用3D Slicer软件,于轴位T2WI和ADC图上,沿肿瘤边缘逐层勾画感兴趣区,生成3D感兴趣容积(volume of interesting,VOI),获得ADC肿瘤全域直方图参数,包括最大值(ADC_(max)),最小值(ADC_(min)),平均值(ADC_(mean)),偏度(skewness),峰度(kurtosis),熵(entropy),第5(5th)、第10(10 th)、第25(25 th)、第50(50 th)、第75(75 th)、第90(90th)和第95(95th)百分位数ADC值。测量肿瘤形态学参数,包括肿瘤体积、肿瘤最大径、矢状位T2WI肿瘤的最大前后径(maximum anteroposterior tumor diameter on sagittal T2-weighted imaging,APsag)和肿瘤面积比(tumor area ratio,TAR)。组内相关系数(intraclass correlation coefficient,ICC)用于评价不同观测者间测量的变异性。利用逻辑回归(logistic regression,LR)构建ADC分数(ADC_(score));联合ADC_(score)、肿瘤的形态学及临床参数构建诺模图,绘制校正曲线及决策曲线。结果:ADC直方图参数中ADC_(min)、ADC_(5th)、ADC_(10th)、ADC_(25th),所有的形态学参数(肿瘤体积、肿瘤最大径、APsag、TAR)和临床参数(年龄)在高级别与低级别EC组间具有统计学差异(P<0.05)。经LR筛选,最终纳入ADC_(min)、ADC_(5th)、ADC_(10th)、ADC_(25th)构建ADC_(score)。联合年龄、APsag及ADC_(score)构建ADC诺模图(经多因素LR分析,以上3个参数为分类高级别与低级别EC的独立风险因素),其预测高级别EC的受试者工作特性曲线下面积(area under curve,AUC)、敏感性、特异性在训练组分别为0.845、81.16%、72.46%,验证组分别为0.842、76.67%、80.00%。校正曲线显示,ADC诺模图预测高级别EC具有较高的准确性;决策曲线显示其预测高级别EC在训练组Objective:To explore and verify the application value of whole tumor volume apparent diffusion coefficient(ADC)nomogram analysis in predicting the high-grade endometrial carcinoma(EC)before operation.Methods:142 patients with EC confirmed by postoperative histopathology were collected retrospectively,and were divided into training cohort(n=99)and validation cohort(n=43)in a ratio of 7∶3.All patients underwent 3.0-T MR examination before operation.Using 3D Slicer software,manually draw ROI along the tumor edge on each layer of image containing tumor margin on axial T2WI and ADC,and accumulate 3D volume of interesting(VOI)and obtain the signal intensity histogram of 3D ROI and its parameters(including the maximum,minimum,mean,skewness,kurtosis,entropy,5th,10th,25th,50th,75th,90th and 95th percentile of ADC values).Measure the tumor morphology parameters,including tumor volume,tumor size,the maximum anteroposterior tumor diameter on sagittal T2-weighted imaging(APsag),and the tumor area ratio(TAR).The intraclass correlation coefficient(ICC)was used to evaluate the variability of measurement.Logistic regression(LR)was used to construct the ADC_(score).The nomogram was constructed by combining the ADC_(score),tumor morphology,and clinical parameters,and the calibration and decision curves were plotted.Results:There were statistically significant differences(P<0.05)in ADC_(min),ADC_(5th),ADC_(10th),ADC_(25th) which were ADC histogram parameters,all morphology parameters(tumor volume,tumor size,APsag,TAR),and clinical parameters(age)between the high-grade and low-grade EC groups.After LR screening,ADC_(min),ADC_(5th),ADC_(10th),and ADC_(25th) were included in the ADC_(score).After binary multifactor LR,age,APsag,and ADC_(score) were finally included as independent risk factors for classifying high-grade and low-grade EC,and the ADC nomogram was constructed.The AUC,sensitivity,and specificity of the ADC nomogram in predicting high-grade EC were 0.845,81.16%,and 72.46%in the training cohort,and 0.842,76.67%,and 80.00%in

关 键 词：子宫内膜癌 磁共振 表观扩散系数 直方图分析 诺模图 病理分级 

分 类 号：R737.33[医药卫生—肿瘤]