计算机辅助系统在诊断乳腺良恶性肿瘤中的应用

目的:探讨计算机辅助诊断系统在良恶性肿瘤检测与特征提取基础上的分类对于乳腺肿瘤的诊断价值。方法:回顾性分析乳腺超声检查发现肿瘤且经过病理学证实的617例患者影像资料，采用手工提取的方式得到乳腺超声图像的感兴趣区域及病灶轮廓，再利用方向梯度直方图（HOG）、局部二值模式（LBP）和灰度共生矩阵（GLCM）3个特征进行乳腺肿瘤的良恶性病变真假阳性检测；最后用受试者操作特征曲线（ROC）分别分析每个特征对于两类病变判别的诊断性能和应用所有特征集合的分类诊断性能。结果:多特征融合方法的各项诊断效能及ROC曲线下面积（AUC）值均优于单特征LBP、HOG、GLCM（P值均<0.05）。与人工诊断相比，多特征融合的敏感性无显著差异，但特异度显著升高达98.57%（Z值=2.25, P<0.05），同时AUC值为0.985，显著优于人工诊断的0.910（Z值=1.99, P<0.05）。结论:计算机辅助系统乳腺超声肿瘤良恶性检测的算法是有效的，能够对乳腺癌鉴别诊断提供有益的参考。

Application of computer-aided detection in ultrasound diagnosis of benign and malignant breast tumors

Objective To explore the diagnostic value of computer-aided detection in breast tumors based on the detection of benign and malignant tumors and feature extraction.Methods The ultrasound images of 617 patients with breast tumors detected by ultrasound and confirmed by pathology were analyzed retrospectively.The regions of interest and lesion contours in the breast ultrasound images were obtained by manual extraction,and then 3 features,namely histogram of oriented gradient(HOG),local binary pattern(LBP)and gray level co-occurrence matrix(GLCM),were used to detect the true or false positive of benign and malignant breast tumors.Finally,receiver operating characteristic(ROC)curve was used to analyze the diagnostic performance of each feature for two types of lesions,and the diagnostic performance of feature set for classification.Results The diagnostic performance and area under ROC curve(AUC)of the detection with the combination of multiple features were superior to those obtained by every single feature(LBP,HOG or GLCM)(all P<0.05).The detection with the combination of multiple features had a sensibility similar to that of manual diagnosis,and a significantly increased specificity which was up to 98.57%(Z value=2.25,P<0.05),and a AUC of 0.985 which was obviously higher than 0.910 of manual diagnosis(Z value=1.99,P<0.05).Conclusion The computer-aided detection for the ultrasound detection of benign and malignant breast tumors is proved to be effective and can provide useful reference for the differentiation and diagnosis of breast tumors.