From the Department of Diagnostic Radiology (D.S.P., K.H., A.M., B.E.D.), Lyda Hill Department of Bioinformatics (S.N., P.K., M.C.C., L.W., A.M.), and Biomedical Engineering Department (A.M.), University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390-8585.
Radiol Imaging Cancer. 2024 May;6(3):e230107. doi: 10.1148/rycan.230107.
Purpose To develop a custom deep convolutional neural network (CNN) for noninvasive prediction of breast cancer nodal metastasis. Materials and Methods This retrospective study included patients with newly diagnosed primary invasive breast cancer with known pathologic (pN) and clinical nodal (cN) status who underwent dynamic contrast-enhanced (DCE) breast MRI at the authors' institution between July 2013 and July 2016. Clinicopathologic data (age, estrogen receptor and human epidermal growth factor 2 status, Ki-67 index, and tumor grade) and cN and pN status were collected. A four-dimensional (4D) CNN model integrating temporal information from dynamic image sets was developed. The convolutional layers learned prognostic image features, which were combined with clinicopathologic measures to predict cN0 versus cN+ and pN0 versus pN+ disease. Performance was assessed with the area under the receiver operating characteristic curve (AUC), with fivefold nested cross-validation. Results Data from 350 female patients (mean age, 51.7 years ± 11.9 [SD]) were analyzed. AUC, sensitivity, and specificity values of the 4D hybrid model were 0.87 (95% CI: 0.83, 0.91), 89% (95% CI: 79%, 93%), and 76% (95% CI: 68%, 88%) for differentiating pN0 versus pN+ and 0.79 (95% CI: 0.76, 0.82), 80% (95% CI: 77%, 84%), and 62% (95% CI: 58%, 67%), respectively, for differentiating cN0 versus cN+. Conclusion The proposed deep learning model using tumor DCE MR images demonstrated high sensitivity in identifying breast cancer lymph node metastasis and shows promise for potential use as a clinical decision support tool. MR Imaging, Breast, Breast Cancer, Breast MRI, Machine Learning, Metastasis, Prognostic Prediction Published under a CC BY 4.0 license.
开发一种定制的深度卷积神经网络（CNN），用于无创预测乳腺癌淋巴结转移。
本回顾性研究纳入了 2013 年 7 月至 2016 年 7 月在作者所在机构接受动态对比增强（DCE）乳腺 MRI 检查的初诊原发性浸润性乳腺癌患者，且患者具有已知的病理（pN）和临床淋巴结（cN）状态。收集临床病理数据（年龄、雌激素受体和人表皮生长因子 2 状态、Ki-67 指数和肿瘤分级）和 cN 和 pN 状态。开发了一个整合动态图像集时间信息的四维（4D）CNN 模型。卷积层学习预测性图像特征，这些特征与临床病理指标相结合，用于预测 cN0 与 cN+和 pN0 与 pN+疾病。使用Receiver Operating Characteristic 曲线（ROC）下面积（AUC）进行评估，采用五重嵌套交叉验证。
对 350 名女性患者（平均年龄 51.7 岁±11.9[标准差]）的数据进行了分析。4D 混合模型区分 pN0 与 pN+的 AUC、敏感性和特异性值分别为 0.87（95%CI：0.83，0.91）、89%（95%CI：79%，93%）和 76%（95%CI：68%，88%），区分 cN0 与 cN+的 AUC、敏感性和特异性值分别为 0.79（95%CI：0.76，0.82）、80%（95%CI：77%，84%）和 62%（95%CI：58%，67%）。
该研究提出的使用肿瘤 DCE MR 图像的深度学习模型在识别乳腺癌淋巴结转移方面具有较高的敏感性，有望成为一种临床决策支持工具。
