应用MRI计算机辅助诊断系统(CAD)评估乳腺小肿块血流动力学特征的价值  

Value of hemodynamic characteristics for small breast masses evaluated by MRI computer-aided diagnosis system

作　　者：宋颖[1] 欧阳汉[1] 叶枫[1] 李静[1] 周纯武[1] 

机构地区：[1]北京协和医学院中国医学科学院肿瘤医院影像诊断科,北京100021

出　　处：《放射学实践》2015年第5期560-565,共6页Radiologic Practice

摘　　要：目的：探讨应用乳腺磁共振计算机辅助系统（MRI-CAD）评价乳腺小肿块血流动力学特征的价值。方法：回顾性分析行MRI检查并经组织学病理证实的肿块型乳腺病灶84个,肿块最大径≤2.0cm。利用CAD系统获得的血流动力学参数分析乳腺小肿块的良恶性特征,包括早期强化峰值、延迟期各曲线类型百分比及单一设定的最可疑恶性曲线类型,并比较CAD法和手动选取感兴趣区法（ROI法）对于鉴别乳腺小肿块良恶性的价值。结果：84个病灶中38个良性病灶,46个恶性病灶。CAD系统计算的良性病变和恶性病变的早期强化峰值平均值分别为（230.54±83.46）%（92%--442%）、（257.72±78.93）%（89%--448%）,两者间差异无统计学意义（P=0.138）。在恶性病灶中流出型、平台型及流入型曲线所占百分比中位数分别为15.15%、20.00%、61.50%,良性病灶中流出型、平台型及流入型曲线所占百分比中位数分别为1.06%、10.08%、87.00%,差异有统计学意义（P〈0.001）。单一设定的最可疑恶性曲线类型表现为流出型病灶共68个,其中良性24个,恶性44个;表现为平台型曲线的病灶共5个,其中良性3个,恶性2个;8个表现为流入型的病灶病理均为良性。当以单一设定的快速流出型曲线为诊断恶性病变的标准时,敏感度为95.7%,特异度为40.0%,ROC曲线下面积（AUC）为0.678（95%CI：0.555--0.801,P=0.006）。CAD系统计算的延迟期流入型曲线所占百分比对于鉴别乳腺小肿块价值最高（AUC=0.798,95%CI：0.701～0.895,P〈0.001）。当流入型曲线所占百分比〈81.50%时,其鉴别良恶性病变的敏感度为89.1%,特异度为60.0%,与ROI法（特异度为42.1%）相比显著提高了诊断乳腺小肿块的特异度（P=0.013）。结论：利用MRI-CAD系统获得的完整病灶延迟期曲线百分比对鉴别乳腺小肿块的良恶性最有价值,与手动选取感兴趣区法相比可显著提高诊断特异度。Objective：To investigate the value of hemodynamic characteristics of small breast masses evaluated by MRI computer-aided diagnosis （CAD） system. Methods：Eighty-four breast mass lesions （longest diameter ≤2.0cm） seen on MRI confirmed by histopathology were reviewed retrospectively. Hemodynamic parameters obtained from CAD system were used for analyzing the benign and malignant features of small breast masses, including initial phase peak enhancement, the percentage of curve type in delayed phase and the most suspicious malignant curve type. The value of two methods （ROI and CAD） for the identification of benign and malignant small breast masses were compared. Results： There were 38 benign and 46 malignant small breast masses found in this study. Initial phase peak enhancement mean values of benign and malig- nant lesions were （230.54±83.46）% （92～442%） and （257.72±78.93）% （89%-448%）,there was no statistically difference between benign and malignant lesions （P= 0. 138）. The percentage median values of three curve type （inflow type,insistent type and outflow type） measured in malignant lesions were 15. 15%, 20. 00%, and 61. 50% respectively while in benign lesions were 1.06 %, 10.08% and 87.00%. Statistical difference was found between benign and malignant lesions （P《0. 001）. In the most suspicious malignant curve type, 68 lesions performed outflow type,including 24 benign and 44 malignant 5 performed insistent type, including 3 benign and 2 malignant all 8 benign lesions performed outflow type. Taking the outflow type as the diagnostic criteria of malignance, the sensitivity was 95.7 %, specificity was 40.0%, the area under cure （AUC） was 0. 678 （95%CI：0. 555--0. 801,P=0. 006）. The percentages of inflow type curve on delayed phase measured by MRI CAD system has the highest accuracy for identifying small breast masses （AUC=0. 798,95% C1： 0. 701--0. 895 ,P〈0. 001）. If percentages less than 81.5% of inflow type curve in a lesion were considered as

关 键 词：乳腺肿瘤 图像处理 计算机辅助 磁共振成像 诊断 鉴别 

分 类 号：R737.9[医药卫生—肿瘤] R445.1[医药卫生—临床医学]