文献速递乳腺US

基于SAFE微波设备与机器学习进行乳腺密度分类

Microwave-Based Breast Density Assessment Using the SAFE Device.
PubMed 文献 · Acad Radiol · 2026-09-11

本研究利用SAFE微波乳腺评估设备采集S11与S21参数的实部和虚部,以机器学习算法分类乳腺密度。在三种模型中,支持向量机表现最好,平均准确率84%、平均敏感度80%、平均特异度86%。结果提示SAFE可望补充现有乳腺密度评估方法。

要点

  • SAFE微波数据可能提供无电离辐射的乳腺密度信息,有助补充X线评估。
  • 支持向量机在本研究中分类性能最优,平均准确率为84%。
  • 该技术目前仍属初步验证,不宜直接替代现有乳腺密度诊断流程。

核心结论

  • 初步结果表明SAFE可被用于乳腺密度分类,并可能补充现有诊断方法。
  • 该结果支持进一步研究微波乳腺密度评估作为临床相关密度信息来源的潜力。

研究设计与样本

  • 研究使用SAFE设备获取微波数据,采用机器学习算法进行乳腺密度分类。
  • 模型输入特征为S11和S21参数的实部与虚部分量。
  • 共评估三种机器学习算法,以标准分类指标评价性能。

诊断标准与关键数值

  • 支持向量机模型平均准确率为84%。
  • 支持向量机模型平均敏感度为80%。
  • 支持向量机模型平均特异度为86%。

临床应用与阅片要点

  • 微波乳腺密度评估可能在X线乳腺密度判断受限时为阅片和报告提供补充信息。
  • 当前结果尚不能替代乳腺密度影像分级,报告中如需引用应注明为实验性技术。

局限性

  • 研究为初步评估,原文未提供样本量、中心数及参考标准等设计信息。
  • 原文仅报告平均性能指标,未说明乳腺密度分级的参考依据及分级类别数量。

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

BACKGROUND: Breast density is a critical factor in both breast cancer risk and the effectiveness of diagnostic technologies. Women with dense breast tissue not only have a higher risk of developing breast cancer but also face challenges in screening and diagnosis using mammography. Microwave-based breast evaluation is an emerging approach that can offer complementary information about breast density without relying on ionizing radiation. Incorporating breast density assessment into microwave-based evaluation may represent a step toward exploring the clinical utility of SAFE, given the established association between breast density and breast cancer risk.

METHODS: We investigated whether data acquired from the SAFE device could be used to classify breast density levels using machine-learning algorithms. The real and imaginary components of the S11 and S21 parameters were used as input features. Multiple machine-learning models were evaluated, and classification performance was assessed using standard metrics.

RESULTS: Among the three machine-learning algorithms evaluated, the best classification performance was achieved using the Support Vector Machine (SVM) model. Results showed promising classification performance with 84% average accuracy, 80% average sensitivity, and 86% average specificity.

CONCLUSION: Our preliminary findings demonstrate that SAFE can be used to classify breast density, potentially complementing existing diagnostic methods. These findings support further investigation of microwave-based breast density assessment as a potentially useful source of clinically relevant breast density information.

来源:Acad Radiol,查看原文。版权归原作者所有。

引用

  1. Karakus, Janjic, Akduman, et al. Microwave-Based Breast Density Assessment Using the SAFE Device. Acad Radiol. 2026. PMID: 42728183. doi:10.1016/j.acra.2026.08.103

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。