联合动态增强磁共振成像 (DCE-MRI) 以及弥散加权成像 (DWI) 的影像特征, 通过建立模型, 分别对乳腺癌的组织学分级以及Ki-67的表达进行预测.对144例未经过任何手术或化疗的乳腺浸润性导管癌患者的数据进行回顾性分析, 这些患者均采用3T扫描仪进行术前乳腺MRI检查, 从中获取DCE-MRI以及DWI影像, 并从DWI中计算得到表观扩散系数 (ADC) .对不同参数磁共振影像进行肿瘤分割, 并分别从整个肿瘤区域中提取纹理特征、统计特征、形态特征等.采用无监督判别特征选择方法 (UDFS) 和Fisher Score算法进行特征选择, 将分类模型分别应用于DCE-MRI及DWI图像数据, 将得到的不同分类器进行多分类器模型融合, 最终得到多参数图像的联合预测结果.为了评估所建立模型的分类性能, 通过留一法交叉验证 (LOOCV) 的方法计算ROC曲线下的面积 (AUC) .对于分级任务, DCE-MRI的第二增强序列达到0.780的最优AUC, (特异度为0.647, 灵敏度为0.934);对于Ki-67预测任务, DWI序列达到0.756的最优AUC (特异度为0.806, 灵敏度为0.695) .经过融合, 分级的预测结果提高到AUC为0.808 (特异度为0.706, 灵敏度为0.895), Ki-67的预测结果提高到AUC为0.783 (特异度为0.778, 灵敏度为0.722) .结果表明, 相比采用单一参数的磁共振图像数据, DCE-MRI和DWI的影像特征联合可以提高分类器的性能.%The purpose of this study was to predict the histological grade of breast cancer and Ki-67 expression using features extracted from dynamic contrast enhancement magnetic resonance imaging (DCE-MRI) and diffusion weighted imaging (DWI). In this study, 144 cases of breast invasive ductal carcinoma were collected, which have not experienced breast surgery or chemotherapy before MRI examination. DW and DCE-MR images were obtained from preoperative breast MRI examination using a 3 T scanner, and ADC map was calculated from DWI. Breast tumor segmentation was performed on all of the image series. After that, image features of texture, statistic, and morphological features of breast tumor were extracted on both the DW and DCE-MR images. The unsupervised discriminative feature selection (UDFS) and Fisher Score algorithm were used for feature selection. The classification model was established on these images respectively, and the classifiers of single-parametric image were fused for prediction. In order to evaluate the classifier performance, the area under the receiver operating characteristic curve (AUC) were calculated in a leave-one-out cross-validation analysis. The predictive model based on the second postcontrast image series of DCE-MRI generated an AUC of 0.780 with the specificity and sensitivity of 0.647 and 0.934 respectively in histological grade task, while in Ki-67 expression task, the model based on DWI generated an AUC of 0.756 with the specificity and sensitivity of 0.806 and 0.695, respectively. After multi-classifier fusion using features both from the DWI and DCE-MRI, the classification result was increased to AUC of 0.808 with the specificity and sensitivity of 0.706 and 0.895 respectively in histological grade prediction task, and generated an AUC of 0.783 with the specificity and sensitivity of 0.778 and 0.722 respectively in Ki-67 expression prediction task. In conclusion, it was showed that compared with each single parametric image alone, the performance of the classifier could be improved by combining features of DCE-MRI and DWI.
展开▼
