College of Medicine and Biological Information Engineering, Northeastern University, 195 Chuangxin Road, Shenyang, China.
Department of Radiology, Affiliated Hospital of Guizhou Medical University, 28 Guiyi Road, Guiyang, China.
Med Phys. 2021 Jun;48(6):2920-2928. doi: 10.1002/mp.14832. Epub 2021 Apr 13.
This research aims to analyze the diagnostic contribution of different discriminative regions of the breast ultrasound image and develop a more effective diagnosis method taking advantage of the discriminative regions' complementarity.
First, the discriminative regions of the original breast ultrasound image as the inner region of the lesion, the marginal zone of the lesion, and the posterior echo region of the lesion were defined. The pretrained Inception-V3 network was used to analyze the diagnostic contribution of these discriminative regions. Then, the network was applied to extract the deep features of the original image and the other three discriminative region images. Since there are many features, principal components analysis (PCA) was used to reduce the dimensionality of the extracted deep features. The selected deep features from different discriminative regions were fused to original image features and sent to the stacking ensemble learning classifier for classification experiments. In this study, 479 cases of breast ultrasound images, including 356 benign lesions and 123 malignant ones, were collected retrospectively and randomly divided into the training and validation set.
Experimental results show that by using Inception-V3, the diagnostic performance of each discriminative region is different, and the diagnostic accuracy and the area under the ROC curve (AUC) of the lesion marginal zone image (78.3%, 0.798) are higher than those of the lesion inner region image (73.3%, 0.763) and the posterior echo region image (71.7%, 0.688), but lower than those of the original image (80.0%, 0.817). Furthermore, the best classification performance was obtained when all the four types of deep features (from the original image and three discriminative region images) were fused, and the ensemble learning for classification evaluation was employed. Compared with the original image, the classification accuracy and AUC increased from 80.83%, 0.818 to 85.00%, 0.872, and the classification sensitivity and specificity varied from 0.710, 0.798 to 0.871, 0.787.
The inner region of the lesion, the marginal zone of the lesion, and the posterior echo region of the lesion play significant roles in the diagnosis of the breast ultrasound image. Deep feature fusion of these three kinds of images and the original image can effectively improve the accuracy of diagnosis.
本研究旨在分析乳腺超声图像不同判别区域的诊断贡献，并利用判别区域的互补性开发更有效的诊断方法。
首先，定义原始乳腺超声图像的判别区域为病变内部区域、病变边缘区域和病变后回声区域。使用预训练的 Inception-V3 网络分析这些判别区域的诊断贡献。然后，将网络应用于提取原始图像和其他三个判别区域图像的深度特征。由于特征较多，采用主成分分析（PCA）对提取的深度特征进行降维。从不同判别区域选择的深度特征与原始图像特征融合，并发送到堆叠集成学习分类器进行分类实验。本研究回顾性收集了 479 例乳腺超声图像，包括 356 例良性病变和 123 例恶性病变，随机分为训练集和验证集。
实验结果表明，使用 Inception-V3 时，每个判别区域的诊断性能不同，病变边缘区域图像（78.3%，0.798）的诊断准确率和 ROC 曲线下面积（AUC）高于病变内部区域图像（73.3%，0.763）和后回声区域图像（71.7%，0.688），但低于原始图像（80.0%，0.817）。此外，当融合原始图像和三个判别区域图像的四种类型的深度特征时，分类性能最佳，同时采用集成学习进行分类评估。与原始图像相比，分类准确率和 AUC 从 80.83%、0.818 提高到 85.00%、0.872，分类敏感性和特异性从 0.710、0.798 提高到 0.871、0.787。
病变内部区域、病变边缘区域和病变后回声区域在乳腺超声图像的诊断中起着重要作用。融合这三种图像和原始图像的深度特征可以有效提高诊断准确率。
