Towards generating more interpretable counterfactuals via concept vectors: a preliminary study on chest X-rays
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912413069082624 |
|---|---|
| author | Maksudov, Bulat Curran, Kathleen Mileo, Alessandra |
| author_facet | Maksudov, Bulat Curran, Kathleen Mileo, Alessandra |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_04058 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Towards generating more interpretable counterfactuals via concept vectors: a preliminary study on chest X-rays Maksudov, Bulat Curran, Kathleen Mileo, Alessandra Image and Video Processing Artificial Intelligence Computer Vision and Pattern Recognition 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. |
| title | Towards generating more interpretable counterfactuals via concept vectors: a preliminary study on chest X-rays |
| topic | Image and Video Processing Artificial Intelligence Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.04058 |