基于定量影像组学的乳腺肿瘤良恶性诊断

乳腺癌是女性致死率最高的恶性肿瘤之一。为提高诊断效率,提供给医生更加客观和准确的诊断结果。借助影像组学的方法,利用公开数据集BreaKHis中82例患者的乳腺肿瘤病理图像,提取乳腺肿瘤病理图像的灰度特征、Haralick纹理特征、局部二值模式(LBP)特征和Gabor特征共139维影像组学特征,并用主成分分析(PCA)对影像组学特征进行降维,然后利用随机森林(RF)、极限学习机(ELM)、支持向量机(SVM)、k最近邻(kNN)等4种不同的分类器构建乳腺肿瘤良恶性的诊断模型,并对上述不同的特征集进行评估。结果表明,基于支持向量机的影像组学特征的分类效果最好,准确率能达到88.2%,灵敏性达到86.62%,特异性达到89.82%。影像组学方法可为乳腺肿瘤良恶性预测提供一种新型的检测手段,使乳腺肿瘤良恶性临床诊断的准确率得到很大提升。

Diagnosis of Benign and Malignant Breast Tumors Using a Quantitative Radiomic Method

Breast cancer is one of the malignant cancers with the highest mortality rate in women. To improve the diagnostic efficiency and provide more objective and accurate diagnosis results, we used a public data set BreaKHis of pathological images of breast tumors in 82 patients by radiomic method. We extracted grayscale features, Haralick texture features, local binary patterns (LBP) features and Gabor features of 139-dimensional radiomic features of breast tumor pathology images from the data set. The principal component analysis (PCA) was employed to reduce the dimensionality of the omics. After that we constructed a diagnostic model of breast tumors by using four different classifiers including random forest (RF), extreme learning machine (ELM), support vector machine (SVM), k-nearest neighbor (kNN) and evaluated the different feature sets mentioned above. Results showed that the classification of radiomics features based on support vector machine was the best. The accuracy rate reached 88.2%, the sensitivity reached 86.62%, and the specificity reached 89.82%. The proposed method provided a new detection solution for the prediction of benign and malignant breast tumors, which would greatly improve the accuracy of clinical diagnosis of benign and malignant breast tumors.