磁共振纹理动态特征鉴别乳腺良恶性肿块

发布时间：2018-04-09 20:14

本文选题：磁共振成像　切入点：纹理　出处：《放射学实践》2017年10期

【摘要】：目的:提取乳腺病灶的时空变化特征作为新的DCE-MRI标记(称为纹理动态特征)并证明其鉴别良恶性肿块的能力。方法:回顾性分析52个乳腺肿块,其中恶性肿瘤30个,良性肿块22个,提取并对动态特征信号强度特征、纹理特征、形态特征、边缘特征进行分组。为了更好评估这些特征,采用不同的特征类建立分组模型,计算正确率,敏感度,特异性及曲线下面积(AUC)。结果:结合纹理动态特征所建立的良恶性肿瘤分类器具有最大的AUC=0.94,准确率90%,敏感度92%,特异性85%,优于其他各组分类器,与信号强度特征所建立的模型差异有统计学意义(AUC=0.80,P0.05)。结论:磁共振纹理动态特征有助于鉴别良恶性肿块,甚至优于临床上最流行的DCE-MRI标记信号强度动态特征。
[Abstract]:Objective: to extract the temporal and spatial characteristics of breast lesions as new DCE-MRI markers (called texture dynamic features) and to demonstrate their ability to differentiate benign and malignant masses.Methods: a retrospective analysis of 52 breast masses, including 30 malignant tumors and 22 benign masses, was carried out. The signal intensity features, texture features, morphological features and edge features of dynamic features were extracted and classified.In order to better evaluate these features, different feature classes are used to establish a grouping model to calculate the accuracy, sensitivity, specificity and area under the curve.Results: the classifier of benign and malignant tumor based on texture dynamic features had the largest AUC0.94, accuracy 90, sensitivity 922, specificity 85, which was superior to other classifiers, and had significant difference from the model established by signal intensity feature.Conclusion: Mr texture dynamic features are helpful in differentiating benign and malignant tumors, even better than the most popular dynamic features of DCE-MRI labeling signal intensity.
【作者单位】：深圳市第二人民医院放射科;
【基金】：深圳市科技研究资金(JCYJ20150330102720117)
【分类号】：R445.2;R737.9

【相似文献】