Liu Yuanyuan, Tong Yunfei, Wan Yun, Xia Ziqiang, Yao Guoyan, Shang Xiaojing, Huang Yan, Chen Lijun, Chen Daniel Q, Liu Bo
Department of Radiology, The Second Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China.
Department of Engineering, Shanghai Yanghe Huajian Artificial Intelligence Technology Co., Ltd, Shanghai, China.
Front Oncol. 2023 Mar 22;13:1119743. doi: 10.3389/fonc.2023.1119743. eCollection 2023.
Architectural distortion (AD) is a common imaging manifestation of breast cancer, but is also seen in benign lesions. This study aimed to construct deep learning models using mask regional convolutional neural network (Mask-RCNN) for AD identification in full-field digital mammography (FFDM) and evaluate the performance of models for malignant AD diagnosis.
This retrospective diagnostic study was conducted at the Second Affiliated Hospital of Guangzhou University of Chinese Medicine between January 2011 and December 2020. Patients with AD in the breast in FFDM were included. Machine learning models for AD identification were developed using the Mask RCNN method. Receiver operating characteristics (ROC) curves, their areas under the curve (AUCs), and recall/sensitivity were used to evaluate the models. Models with the highest AUCs were selected for malignant AD diagnosis.
A total of 349 AD patients (190 with malignant AD) were enrolled. EfficientNetV2, EfficientNetV1, ResNext, and ResNet were developed for AD identification, with AUCs of 0.89, 0.87, 0.81 and 0.79. The AUC of EfficientNetV2 was significantly higher than EfficientNetV1 (0.89 vs. 0.78, P=0.001) for malignant AD diagnosis, and the recall/sensitivity of the EfficientNetV2 model was 0.93.
The Mask-RCNN-based EfficientNetV2 model has a good diagnostic value for malignant AD.
结构扭曲（AD）是乳腺癌常见的影像学表现，但也可见于良性病变。本研究旨在构建基于掩膜区域卷积神经网络（Mask-RCNN）的深度学习模型，用于全视野数字乳腺摄影（FFDM）中AD的识别，并评估模型对恶性AD诊断的性能。
本回顾性诊断研究于2011年1月至2020年12月在广州中医药大学第二附属医院进行。纳入FFDM中乳腺存在AD的患者。采用Mask RCNN方法开发用于AD识别的机器学习模型。采用受试者工作特征（ROC）曲线、曲线下面积（AUC）及召回率/敏感度评估模型。选择AUC最高的模型进行恶性AD诊断。
共纳入349例AD患者（190例为恶性AD）。开发了EfficientNetV2、EfficientNetV1、ResNext和ResNet用于AD识别，AUC分别为0.89、0.87、0.81和0.79。在恶性AD诊断中，EfficientNetV2的AUC显著高于EfficientNetV1（0.89对0.78，P=0.001），且EfficientNetV2模型的召回率/敏感度为0.93。
基于Mask-RCNN的EfficientNetV2模型对恶性AD具有良好的诊断价值。
