用概念向量生成可解释的假想影像，助力医学影像模型可信化。
Towards generating more interpretable counterfactuals via concept vectors: a preliminary study on chest X-rays
- 通过自编码器将临床概念映射到生成模型隐空间，提取概念激活向量。
- 在胸片上生成夸大或消除特定病灶的反事实图像，对大病灶效果显著。
- 适合医疗AI可解释性研究者，尤其关注临床知识对齐的场景。
医疗影像模型部署的关键挑战在于与临床知识对齐并具备可解释性。本文将临床概念映射至生成模型的隐空间，提取概念激活向量（CAVs）。采用简单的重构自编码器，无需显式标签训练即可将用户定义的概念关联到图像级特征。所提取的概念在不同数据集间具有稳定性，可生成突出临床相关特征的可视化解释。沿概念方向遍历隐空间，生成夸大或减少特定临床特征的反事实图像。在胸片上的初步实验显示，对心影增大等大病灶具有良好效果，而小病灶受限于重建能力仍具挑战。尽管未超越基线性能，该方法为实现与临床知识一致的、基于概念的可解释性提供了可行路径。
原文摘要 · Abstract (English)
An essential step in deploying medical imaging models is ensuring alignment with clinical knowledge and interpretability. We focus on mapping clinical concepts into the latent space of generative models to identify Concept Activation Vectors (CAVs). Using a simple reconstruction autoencoder, we link user-defined concepts to image-level features without explicit label training. The extracted concepts are stable across datasets, enabling visual explanations that highlight clinically relevant features. By traversing latent space along concept directions, we produce counterfactuals that exaggerate or reduce specific clinical features. Preliminary results on chest X-rays show promise for large pathologies like cardiomegaly, while smaller pathologies remain challenging due to reconstruction limits. Although not outperforming baselines, this approach offers a path toward interpretable, concept-based explanations aligned with clinical knowledge.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。