文献速递乳腺乳腺X线

基于乳腺X线摄影合成数字乳腺断层摄影特征的深度学习用于乳腺癌诊断

Deep learning synthesis of DBT features from mammography for breast cancer diagnosis.
PubMed 文献 · Eur J Radiol · 2025-10-23

本回顾性研究纳入806例样本(训练集570例、测试集236例),构建跨模态特征合成(CMFS)模型,从数字乳腺摄影(DM)图像生成合成数字乳腺断层摄影(sDBT)特征。基于DBT的模型AUC达0.916;利用DM生成的sDBT特征使仅依赖DM数据的预测模型AUC从0.845提升至0.878(p=0.002)。

要点

  • 基于DBT的模型AUC为0.916,采用32通道注意力偏置与MSAN架构。
  • DM生成的sDBT特征使仅用DM的模型AUC从0.845提升至0.878(p=0.002)。
  • 模型可视化显示病灶定位精确,可能降低患者辐射暴露。

核心结论

  • 所提出的CMFS框架通过从DM生成sDBT特征改善乳腺癌诊断,在提高诊断准确性的同时降低辐射暴露。

研究设计与样本

  • 回顾性研究,经机构审查委员会批准,2019年2月至2020年8月连续收集样本,共806例(训练集570例、测试集236例),开发跨模态特征合成(CMFS)模型,采用基于ResNet的多尺度注意力网络(MSAN)与生成对抗网络(GAN)多任务学习。

诊断标准与关键数值

  • 基于DBT的模型AUC为0.916,采用32通道注意力偏置与MSAN架构。
  • 仅依赖DM数据的预测模型AUC为0.845,使用DM生成的sDBT特征后AUC提升至0.878(p=0.002)。

临床应用与阅片要点

  • 模型可视化显示病灶定位精确。
  • 该方法利用DBT数据增强DM的诊断价值,可能降低患者辐射暴露。

局限性

  • 原文未报告局限性

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

OBJECTIVES: Digital breast tomosynthesis (DBT) enhances diagnostic accuracy by minimizing tissue overlap seen in digital mammography (DM). However, the substantial number of images generated by DBT poses challenges for radiologists. The aim of this study was to enhance the diagnostic accuracy and clinical utility of DM for breast cancer diagnosis by leveraging advanced feature representations derived from DBT.

MATERIAL AND METHODS: This retrospective study was approved by the institutional review board of the institute, and samples were consecutively collected from February 2019 to August 2020. Our dataset comprised 806 samples (training: n = 570; testing: n = 236). We developed a cross-modality feature synthesis (CMFS) model to generate synthetic DBT (sDBT) features from DM images. To adeptly capture and analyze structures of varying sizes, such as tumors and their surrounding glandular regions, we designed a ResNet-based multiscale attention network (MSAN). Our generative adversarial network (GAN) model employs multitask learning to convert mammography features into sDBT features for breast cancer diagnosis.

RESULTS: Our DBT-based model achieved an area under the curve (AUC) of 0.916, utilizing a 32-channel attention bias and the MSAN architecture. Furthermore, the prediction model using the sDBT features generated from DM enhanced lesion classification, improving the AUC from 0.845 to 0.878 (p = 0.002) when relying solely on DM data. Ablation studies validated the crucial components of our model, underscoring its efficacy in breast cancer diagnosis. Model visualizations demonstrated precise lesion localization, and our method enhances the diagnostic value of DM by utilizing DBT data, potentially reducing patient radiation exposure.

CONCLUSIONS: The proposed CMFS framework improves breast cancer diagnosis by generating sDBT features from DM, enhancing diagnostic accuracy while reducing radiation exposure. Code and models are available at https://github.com/ibml2024/CMFS.

来源:Eur J Radiol,查看原文。版权归原作者所有。

引用

  1. Fan M, Wang L, Wang J, et al. Deep learning synthesis of DBT features from mammography for breast cancer diagnosis. Eur J Radiol. 2025. PMID: 41151435. doi:10.1016/j.ejrad.2025.112486

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。