文献速递乳腺甲状腺与浅表妇产影像US

多尺度CNN Token Transformer用于可解释超声图像分类

MSCT-Trans: A Multi-scale Convolutional Neural Network Token Transformer for Interpretable Ultrasound Image Classification.
PubMed 文献 · Ultrasound Med Biol · 2026-08-25

提出轻量级混合架构MSCT-Trans,结合CNN多尺度特征与Transformer全局建模,用于乳腺、甲状腺及胎儿超声图像分类。在三个超声基准数据集上,该模型在准确率、macro-F1及AUC方面均优于CNN和Transformer基线。可解释性分析显示其能关注诊断相关解剖区域,并产生校准良好的置信度估计。

要点

  • MSCT-Trans在类别不平衡及数据有限条件下仍优于CNN和Transformer基线。
  • 模型注意力聚焦于诊断相关解剖区域,有助于提升临床透明度。
  • 预测错误与模型不确定性相关,而非过度自信的误分类。

核心结论

  • MSCT-Trans在乳腺、甲状腺及胎儿超声基准数据集上持续优于CNN和Transformer基线,尤其在类别不平衡和有限数据条件下。

研究设计与样本

  • 提出多尺度CNN Token Transformer混合架构,使用预训练CNN骨干提取多尺度特征图并转换为统一token序列。
  • 在三个超声基准数据集上评估:乳腺(BUS-BRA+BUSI+UCLM)、甲状腺(TN5000)及胎儿成像。
  • 进行两部分可解释性分析:Grad-CAM++空间定位与softmax类别概率分解。

关键影像表现/征象

  • 模型注意力持续集中于诊断相关的解剖区域。
  • 产生校准良好的置信度估计,预测错误与模型不确定性相关联,而非过度自信的误分类。

诊断标准与关键数值

  • 以准确率、macro-F1及AUC为评价指标,MSCT-Trans均优于CNN和Transformer基线。

临床应用与阅片要点

  • 该架构可作为基于超声的临床决策支持的透明且可信的基础模型。
  • 模型输出的空间定位图可用于辅助判断模型是否关注了正确的解剖结构。

局限性

  • 原文仅报告在公开基准数据集上的性能,未报告外部临床验证或前瞻性研究结果。

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

Automated ultrasound image classification is increasingly important for clinical decision support in breast, thyroid and fetal screening. However, deploying deep learning models in such safety-critical settings demands not only high predictive accuracy but also transparency, interpretability and trustworthiness-properties that existing approaches address insufficiently. Convolutional neural networks (CNNs) capture local texture patterns but struggle with global contextual dependencies, while Transformer-based models offer long-range reasoning yet require large-scale training data and remain sensitive to ultrasound-specific noise, both limiting factors for clinical deployment. We propose Multi-Scale CNN Token Transformer (MSCT-Trans), a lightweight and interpretable hybrid architecture for general-purpose ultrasound image classification. MSCT-Trans extracts multi-scale feature maps from a pre-trained CNN backbone and converts them into a unified token sequence, enabling a Transformer encoder to model global dependencies and inter-scale interactions over semantically meaningful, noise-attenuated representations. To support clinical transparency, we conducted a two-part explainability analysis-Gradient-weighted Class Activation Mapping++ spatial localisation and softmax class probability breakdown-demonstrating that MSCT-Trans consistently attends to diagnostically relevant anatomical regions, produces well-calibrated confidence estimates and associates prediction errors with model uncertainty rather than over-confident mis-classification. Here we evaluated MSCT-Trans on three ultrasound benchmarks spanning breast (BUS-BRA + BUSI + UCLM), thyroid (TN5000) and fetal imaging. MSCT-Trans consistently outperformed CNN and Transformer baselines across accuracy, macro-F1 and area under the receiver operating characteristic curve, particularly under class imbalance and limited data regimens. The combination of strong predictive performance, spatially grounded interpretability and calibrated uncertainty estimation positions MSCT-Trans as a transparent and trustworthy foundation for ultrasound-based clinical decision support. Code: https://github.com/MohsinFurkh/MSCT-Trans.

来源:Ultrasound Med Biol,查看原文。版权归原作者所有。

引用

  1. Dar, Mukhtar. MSCT-Trans: A Multi-scale Convolutional Neural Network Token Transformer for Interpretable Ultrasound Image Classification. Ultrasound Med Biol. 2026. PMID: 42637640. doi:10.1016/j.ultrasmedbio.2026.07.032

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。