Zhang Qian, Lou Yang, Liu Xiaofeng, Liu Chong, Ma Wenjuan
Department of Medical Imaging, The First Central Hospital of Baoding, Baoding, China.
Department of Radiology, Affiliated Hospital of Hebei University, Baoding, China.
Gland Surg. 2025 Feb 28;14(2):228-237. doi: 10.21037/gs-2024-495. Epub 2025 Feb 25.
Accurate evaluation of the axillary lymph node (ALN) status is needed for determining the treatment protocol for breast cancer. This study aimed to build an artificial intelligence (AI) model to predict ALN metastasis based on pre-treatment dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI) of breast cancer combined with radiomics algorithms.
Pre-treatment DCE-MRI dataset of 166 patients with pathologically confirmed breast cancer diagnosis from January 2017 to August 2020 was collected, and all patients were randomly divided into a training group and test group with a ratio of 7:3. Each patient underwent pre-enhancement as well as post-enhancement 1-6 MRI, and a total of 7,224 two-dimensional (2D) and 9,863 three-dimensional (3D) features were extracted, respectively. Radiomics models based on 2D, 3D, pre-enhancement, and the first post-enhancement images were established using the least absolute shrinkage selection operator (LASSO) algorithm based on machine learning, and the area under the curve (AUC), accuracy, sensitivity, and specificity of the models were calculated.
The mean AUC, accuracy, sensitivity, and specificity of the 10-fold cross-validation of the 3D radiomics-based model were 82%, 82%, 83%, and 81%, respectively. The C-index of the combined model with combining radiomics features and clinical features was 90%, the AUC was 90%, the specificity was 91%, the sensitivity was 77% and the accuracy was 84%.
The comprehensive prediction model using DCE-MRI image combined with clinical features can accurately predict ALN metastasis in breast cancer.
准确评估腋窝淋巴结（ALN）状态对于确定乳腺癌的治疗方案至关重要。本研究旨在构建一种人工智能（AI）模型，基于乳腺癌治疗前动态对比增强磁共振成像（DCE-MRI）结合放射组学算法来预测ALN转移。
收集了2017年1月至2020年8月期间166例经病理确诊为乳腺癌的患者的治疗前DCE-MRI数据集，所有患者以7:3的比例随机分为训练组和测试组。每位患者均接受了增强前及增强后1-6次MRI检查，分别提取了总共7224个二维（2D）特征和9863个三维（3D）特征。基于机器学习的最小绝对收缩选择算子（LASSO）算法，建立了基于2D、3D、增强前及首次增强后图像的放射组学模型，并计算了模型的曲线下面积（AUC）、准确率、敏感性和特异性。
基于3D放射组学模型的10倍交叉验证的平均AUC、准确率、敏感性和特异性分别为82%、82%、83%和81%。结合放射组学特征和临床特征的联合模型的C指数为90%，AUC为90%，特异性为91%，敏感性为77%，准确率为84%。
使用DCE-MRI图像结合临床特征的综合预测模型能够准确预测乳腺癌中的ALN转移。
