磁共振全域直方图分析鉴别子宫富于细胞平滑肌瘤和肌瘤变性的价值  

The value of whole-lesion histogram analysis of MR images in differentiating uterine cellular leiomyoma from degeneration of uterine leiomyoma

作　　者：方如旗[1] 董梦雅 杨青霞 陈霞平[1] 翁淑萍[1] 周作福[1] FANG Ruqi;DONG Mengya;YANG Qingxia;CHEN Xiaping;WENG Shuping;ZHOU Zuofu(Department of Radiology,Fujian Maternity and Child Health Hospital,Fuzhou 350000,China;Department of Radiology,Fujian Provincial Children's Hospital,Fuzhou 350000,China)

机构地区：[1]福建省妇幼保健院影像科,福州350000 [2]福建省儿童医院放射科,福州350000

出　　处：《磁共振成像》2021年第11期66-69,79,共5页Chinese Journal of Magnetic Resonance Imaging

基　　金：福建省卫生健康科研人才培养项目(中青年骨干人才培养项目)(编号:2019-ZQN-26)。

摘　　要：目的评估肿瘤磁共振全域直方图分析鉴别子宫富于细胞平滑肌瘤(uterine cellular leiomyoma,UCL)和子宫肌瘤变性(degeneration of uterine leiomyoma,UL-D)的价值。材料与方法回顾性研究2016年3月至2021年4月经病理证实的48例子宫富于细胞平滑肌瘤和44例UL-D患者的影像和临床资料,术前均行1.5 T磁共振常规扫描和扩散加权成像(diffusion weighted imaging,DWI)扫描,2名医师利用MaZda软件在T2WI(T2 weighted imaging)及ADC图上独立勾画肿瘤感兴趣区(volume of interest,VOI),进行灰度图直方图分析,提取T2WI及ADC图灰度直方图参数,包括均值(mean)、方差(variance)、偏度(skewness)、峰度(kurtosis)、第1、10、50、90、99百分位数(Perc 1%、Perc 10%、Perc 50%、Perc 90%、Perc 99%)。采用组内相关系数(intra-class correlation coefficients,ICC)评价不同观察者判断的一致性,利用二元Logistic回归方法筛选参数以构建预测模型,利用ROC曲线分析评估单一参数及Logistic回归模型诊断效能。结果组内ICC为0.844,两组间ADC直方图参数mean、variance、skewness、Perc 1%、Perc 10%、Perc 50%、Perc 90%、Perc 99%及T2WI直方图参数variance、kurtosis、Perc 1%有统计学意义(P<0.05),绘制其受试者工作特征(ROC)曲线,曲线下面积(AUC)分别为0.833、0.677、0.674、0.736、0.777、0.824、0.848、0.822及0.705、0.660、0.640。二元Logistic回归分析及ROC分析,T2WI、ADC、T2WI+ADC模型AUC分别为0.790、0.848、0.881,相对应约登指数分别为0.4830、0.6250、0.6288。ROC曲线两两比较,T2WI与ADC、(T2WI+ADC)与T2WI、(T2WI+ADC)与ADC的P值分别为0.3425、0.0394、0.2348。结论肿瘤全域(T2WI、ADC)直方图定量参数值有助于子宫富于细胞平滑肌瘤与UL-D的鉴别诊断,联合T2WI+ADC模型具有最高的曲线下面积,但诊断效能不优于ADC参数。Objectives:To assess the utility of whole-lesion histogram of multiple MRI sequences for differentiating uterine cellular leiomyoma(UCL)from degeneration of uterine leiomyoma(UL-D).Materials and Methods:Fourty eight patients with UCL and fourty four patients with UL-D from March 2016 to April 2021,who underwent preoperative routine pelvic MRI sequences and diffusion weighted imaging(DWI),were retrospectively evaluated.Two experienced radiologists manually delineated the volume of interest(VOI)by MaZda package in T2 weighted imaging(T2WI)and ADC images,and the values derived from whole-lesion histogram analysis(including mean,variance,skewness,Perc 1%,Perc 10%,Perc 50%,Perc 90%,Perc 99%)were measured for each volume of interest(VOI).The consistency of assessment between the two radiologists was evaluated by using intra-class correlation coefficients(ICC),the MR variables were selected to build Logistic regression model.Receiver operating characteristic(ROC)curve analysis was performed to evaluate the diagnostic performance of single variable and Logistic regression model to differentiate UCL from UL-D.Results:The intra-class correlation coefficients between radiologists was 0.844,the features of mean,variance,skewness,Perc 1%,Perc 10%,Perc 50%,Perc 90%,Perc 99%extracted from ADC maps and features of variance,kurtosis,Perc 1%extracted from T2WI images showed statistically significant(P<0.05).The area under the curve(AUC)were 0.833,0.677,0.674,0.736,0.777,0.824,0.848,0.822 and 0.705,0.660,0.640 for each feature respectively.The T2WI,ADC,T2WI combined ADC Logistic regression models were built to differentiate UCL from UL-D,created AUCs of 0.790,0.848,0.881 respectively,with corresponding Youden's index were 0.4830,0.6250,0.6288.Pairwise comparison of ROC curves of each model,the P values were 0.3425,0.0394,0.2348 for T2WI vs.ADC,(T2WI+ADC)vs.T2WI,(T2WI+ADC)vs.ADC respectively.Conclusions:Whole-lesion histogram analysis of T2WI and ADC gray-level images may assist in differentitating between UCL and UL-D patients.The

关 键 词：子宫平滑肌瘤 子宫肌瘤变性 富于细胞 表观扩散系数 直方图 磁共振成像 

分 类 号：R445.2[医药卫生—影像医学与核医学] R737.33[医药卫生—诊断学]