根据磁共振T2加权影像特征预测乳腺癌组织学分级

组织学分级是乳腺癌的重要预后指标。探讨磁共振(MRI)T2加权影像特征与乳腺癌组织学分级的关联性, 可为术前预测乳腺癌组织学分级提供有意义的参考作用。回顾性分析167例术前行MRI检查并经病理诊断为浸润性乳腺癌患者, 其中组织学分级Ⅱ级和Ⅲ级的分别为72例和95例。利用计算机半自动方法, 分割患者乳腺磁共振T2加权影像的病灶区域, 并对其提取包括纹理特征和形态特征的40维影像特征。采用留一法交叉验证方法(LOOCV), 通过统计学方法分析比较影像特征在组织分级Ⅱ级和Ⅲ级组间的差异, 并设计多变量分类预测模型。绘制受试者工作特征曲线(ROC), 并计算对应曲线下的面积(AUC);计算敏感性、特异性、F1-Measure等指标, 对预测模型进行综合评估。对每一维特征进行单变量逻辑回归分析, 在低分级和高分级组间进行统计检验分析(t检验)。形态特征中最优单特征为病灶半径, AUC值为0.742(P<0.05);纹理特征中最优特征为大面积高灰度级增强, AUC值为0.742(P<0.05)。设计多元逻辑回归(MLR)、支持向量机(SVM)、多任务学习(MTL)等3种分类器, 预测乳腺癌组织学分级, 其AUC值分别为0.767±0.036、0.772±0.036和0.771±0.037, 对应特异性分别为0.667、0.653、0.708, 灵敏度分别为0.747、0.737、0.684。研究表明, 乳腺癌的T2加权影像特征在一定程度上反映其组织学分级, 对乳腺癌的预后判断具有潜在价值。

Prediction of Histological Grade in Invasive Breast Cancer Based on T2-Weighted MRI

The purpose of this study was to predict the histological grade of invasive breast cancer based on radiomic analysis of T2-weighted magnetic resonance images (MRI). A dataset of 167 invasive breast cancer cases who had preoperative breast MRI with a 3.0 T scanner were collected. Among them, 95 cases were diagnosed as high-grade malignant (Grade 3) invasive breast cancer, while 72 were mediate-grade malignant (Grade 2). Semi-automatic lesion segmentation was performed on each T2-weighted MRI, in which 30 texture and 10 morphological features were extracted. A univariate logistic regression classifier model was implemented to evaluate the performance of the individual feature for discriminating histological grade. Various classifiers including multivariate logistic regression (MLR), support vector machines (SVM) and multi-task learning (MTL) were utilized and compared for classification. The diagnostic performance was evaluated by the area under the curve (AUC) with the receiver operating characteristic (ROC) analysis under leave-one-out cross-validation (LOOCV). P-value was calculated using Student's t test. The best single feature of morphology was the lesion radius, which was the AUC value of 0.742 and P value of 0.749×10^-9. The best-performance texture feature was large zone high gray emphasis, with the AUC value of 0.742 and theP value of 0.175×10^-3. The AUC values from classifiers of MLR, SVM and MTL were 0.767±0.036, 0.772±0.036 and 0.771±0.037, respectively. The values of specificity were 0.667, 0.653 and 0.708, respectively while the values of sensitivity were 0.747, 0.737 and 0.684, respectively. The results showed that T2-weighted MRI features could be utilized as promising biomarkers for predicting histological grade in invasive breast cancer.