From the Sino-German Tongji-Caritas Research Center of Ultrasound in Medicine, Department of Medical Ultrasound, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, Hubei Province, China (L.Q.Z., G.G.W., Q.W., Y.B.D., X.W.C., C.F.D.); School of Mathematics and Computer Science, Wuhan Textile University, Wuhan, Hubei Province, China (X.L.W.); Department of Ultrasound, The First People's Hospital of Huaihua, University of South China, Huaihua, China (S.Y.H.); Department of Ultrasound, China Resources & Wisco General Hospital, Wuhan, Hubei Province, China (H.R.Y.); Department of Ultrasound, Affiliated Hangzhou First People's Hospital, Zhejiang University School of Medicine, Hangzhou, Zhejiang, China (L.Y.B.); Department of Thyroid and Breast Surgery, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, Hubei Province, China (X.R.L.); and Medical Clinic 2, Caritas-Krankenhaus Bad Mergentheim, Academic Teaching Hospital of the University of Wuerzburg, Bad Mergentheim, Germany (C.F.D.).
Radiology. 2020 Jan;294(1):19-28. doi: 10.1148/radiol.2019190372. Epub 2019 Nov 19.
Background Deep learning (DL) algorithms are gaining extensive attention for their excellent performance in image recognition tasks. DL models can automatically make a quantitative assessment of complex medical image characteristics and achieve increased accuracy in diagnosis with higher efficiency. Purpose To determine the feasibility of using a DL approach to predict clinically negative axillary lymph node metastasis from US images in patients with primary breast cancer. Materials and Methods A data set of US images in patients with primary breast cancer with clinically negative axillary lymph nodes from Tongji Hospital (974 imaging studies from 2016 to 2018, 756 patients) and an independent test set from Hubei Cancer Hospital (81 imaging studies from 2018 to 2019, 78 patients) were collected. Axillary lymph node status was confirmed with pathologic examination. Three different convolutional neural networks (CNNs) of Inception V3, Inception-ResNet V2, and ResNet-101 architectures were trained on 90% of the Tongji Hospital data set and tested on the remaining 10%, as well as on the independent test set. The performance of the models was compared with that of five radiologists. The models' performance was analyzed in terms of accuracy, sensitivity, specificity, receiver operating characteristic curves, areas under the receiver operating characteristic curve (AUCs), and heat maps. Results The best-performing CNN model, Inception V3, achieved an AUC of 0.89 (95% confidence interval [CI]: 0.83, 0.95) in the prediction of the final clinical diagnosis of axillary lymph node metastasis in the independent test set. The model achieved 85% sensitivity (35 of 41 images; 95% CI: 70%, 94%) and 73% specificity (29 of 40 images; 95% CI: 56%, 85%), and the radiologists achieved 73% sensitivity (30 of 41 images; 95% CI: 57%, 85%; = .17) and 63% specificity (25 of 40 images; 95% CI: 46%, 77%; = .34). Conclusion Using US images from patients with primary breast cancer, deep learning models can effectively predict clinically negative axillary lymph node metastasis. Artificial intelligence may provide an early diagnostic strategy for lymph node metastasis in patients with breast cancer with clinically negative lymph nodes. Published under a CC BY 4.0 license. See also the editorial by Bae in this issue.
背景 深度学习（DL）算法在图像识别任务中表现出色，因此受到广泛关注。DL 模型可以自动对复杂的医学图像特征进行定量评估，并以更高的效率实现更准确的诊断。目的 确定使用深度学习方法从原发性乳腺癌患者的超声图像预测临床阴性腋窝淋巴结转移的可行性。材料与方法 本研究收集了来自同济医院（2016 年至 2018 年共 974 项研究，756 例患者）和湖北省肿瘤医院（2018 年至 2019 年共 81 项研究，78 例患者）的原发性乳腺癌患者具有临床阴性腋窝淋巴结的超声图像数据。腋窝淋巴结状态通过病理检查证实。在同济医院数据集中的 90%数据上训练 3 种不同的卷积神经网络（CNN），即 Inception V3、Inception-ResNet V2 和 ResNet-101 架构，在剩余的 10%数据上进行测试，并在独立测试集中进行测试。将模型的性能与 5 名放射科医生进行比较。从准确性、敏感性、特异性、受试者工作特征曲线、受试者工作特征曲线下面积（AUC）和热图方面分析模型的性能。结果 在独立测试集中，表现最佳的 CNN 模型 Inception V3 对腋窝淋巴结转移的最终临床诊断的预测获得了 0.89 的 AUC（95%置信区间 [CI]：0.83，0.95）。该模型的敏感性为 85%（41 张图像中的 35 张；95%CI：70%，94%），特异性为 73%（40 张图像中的 29 张；95%CI：56%，85%），放射科医生的敏感性为 73%（41 张图像中的 30 张；95%CI：57%，85%； =.17），特异性为 63%（40 张图像中的 25 张；95%CI：46%，77%； =.34）。结论 使用原发性乳腺癌患者的超声图像，深度学习模型可以有效预测临床阴性腋窝淋巴结转移。人工智能可能为临床阴性淋巴结的乳腺癌患者提供淋巴结转移的早期诊断策略。在 CC BY 4.0 许可下发布。 另见本期 Bae 的社论。
Zotero 插件