Wang Xiaoxia, Hu Xiaofei, Wang Churan, Yang Hua, Hu Yan, Lan Xiaosong, Huang Yao, Cao Ying, Yan Lijun, Zhang Fandong, Yu Yizhou, Zhang Jiuquan
Department of Radiology, Chongqing University Cancer Hospital, Chongqing Key Laboratory for Intelligent Oncology in Breast Cancer (iCQBC), No. 181 Hanyu Road, Shapingba District, Chongqing, China 400030.
Department of Nuclear Medicine, Southwest Hospital, Third Military Medical University (Army Medical University), Chongqing, China.
Radiol Imaging Cancer. 2025 May;7(3):e240184. doi: 10.1148/rycan.240184.
Purpose To build a deep learning framework using contrast-enhanced MRI for lesion segmentation and automatic molecular subtype classification in breast cancer. Materials and Methods This retrospective multicenter study included patients with biopsy-proven invasive breast cancer between January 2015 and January 2021. An automatic breast lesion segmentation model was developed using three-dimensional (3D) ResU-Net as the backbone, and its accuracy was evaluated in an internal and two external testing datasets using the Dice score. An ensemble model for classification of breast cancer into four molecular subtypes (Ensemble ResNet) was then developed by combining both two-dimensional and 3D lesion features. The performance of Ensemble ResNet was evaluated in the three testing datasets using the area under the receiver operating characteristic curve (AUC). Results A total of 687 female patients (mean age ± SD, 48.70 years ± 8.97) were included, with 289, 61, 73, and 264 patients included in the training, internal testing, and two external testing datasets, respectively. The proposed segmentation model achieved high accuracy in internal testing dataset 1, external testing dataset 2, and external testing dataset 3 (Dice scores: 0.86, 0.82, 0.85) and luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)-enriched, and triple-negative breast cancer (TNBC) subtypes (Dice scores: 0.8571, 0.8323, 0.8199, 0.8481). Ensemble ResNet demonstrated high performance for the prediction of luminal A subtypes (AUC range, 0.74-0.84), luminal B subtypes (AUC range, 0.68-0.72), HER2-enriched subtypes (AUC range, 0.73-0.82), and TNBC (AUC range, 0.80-0.81) in the three testing datasets. Conclusion The proposed novel deep learning framework based on MRI achieved high, robust performance in fully automatic classification of breast cancer molecular subtypes. MR-Imaging, Breast, Oncology, Breast Cancer, Molecular Subtype, Deep Learning Framework © RSNA, 2025.
目的 构建一个使用对比增强磁共振成像（MRI）的深度学习框架，用于乳腺癌病变分割和自动分子亚型分类。材料与方法 这项回顾性多中心研究纳入了2015年1月至2021年1月间经活检证实为浸润性乳腺癌的患者。使用三维（3D）ResU-Net作为主干构建了一个自动乳腺病变分割模型，并使用Dice分数在一个内部测试数据集和两个外部测试数据集中评估其准确性。然后，通过结合二维和3D病变特征，开发了一个将乳腺癌分为四种分子亚型的集成模型（集成ResNet）。使用受试者操作特征曲线（ROC）下面积（AUC）在三个测试数据集中评估集成ResNet的性能。结果 共纳入687例女性患者（平均年龄±标准差，48.70岁±8.97），其中分别有289例、61例、73例和264例患者纳入训练、内部测试和两个外部测试数据集。所提出的分割模型在内部测试数据集1、外部测试数据集2和外部测试数据集3中取得了较高的准确性（Dice分数：0.86、0.82、0.85），以及在管腔A型、管腔B型、人表皮生长因子受体2（HER2）富集型和三阴性乳腺癌（TNBC）亚型中（Dice分数：0.8571、0.8323、0.8199、0.8481）。集成ResNet在三个测试数据集中对管腔A型亚型（AUC范围，0.74 - 0.84）、管腔B型亚型（AUC范围，0.68 - 0.72）、HER2富集型亚型（AUC范围，0.73 - 0.82）和TNBC（AUC范围，0.80 - 0.81）的预测表现出高性能。结论 所提出的基于MRI的新型深度学习框架在乳腺癌分子亚型的全自动分类中取得了高且稳健的性能。磁共振成像、乳腺、肿瘤学、乳腺癌、分子亚型、深度学习框架 © RSNA，2025年
