{"ID":2857091,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.10174","arxiv_id":"2510.10174","title":"ViConEx-Med: Visual Concept Explainability via Multi-Concept Token Transformer for Medical Image Analysis","abstract":"Concept-based models aim to explain model decisions with human-understandable concepts. However, most existing approaches treat concepts as numerical attributes, without providing complementary visual explanations that could localize the predicted concepts. This limits their utility in real-world applications and particularly in high-stakes scenarios, such as medical use-cases. This paper proposes ViConEx-Med, a novel transformer-based framework for visual concept explainability, which introduces multi-concept learnable tokens to jointly predict and localize visual concepts. By leveraging specialized attention layers for processing visual and text-based concept tokens, our method produces concept-level localization maps while maintaining high predictive accuracy. Experiments on both synthetic and real-world medical datasets demonstrate that ViConEx-Med outperforms prior concept-based models and achieves competitive performance with black-box models in terms of both concept detection and localization precision. Our results suggest a promising direction for building inherently interpretable models grounded in visual concepts. Code is publicly available at https://github.com/CristianoPatricio/viconex-med.","short_abstract":"Concept-based models aim to explain model decisions with human-understandable concepts. However, most existing approaches treat concepts as numerical attributes, without providing complementary visual explanations that could localize the predicted concepts. This limits their utility in real-world applications and parti...","url_abs":"https://arxiv.org/abs/2510.10174","url_pdf":"https://arxiv.org/pdf/2510.10174v1","authors":"[\"Cristiano Patrício\",\"Luís F. Teixeira\",\"João C. Neves\"]","published":"2025-10-11T11:24:47Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Transformer\"]","has_code":false,"code_links":[{"ID":608412,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2857091,"paper_url":"https://arxiv.org/abs/2510.10174","paper_title":"ViConEx-Med: Visual Concept Explainability via Multi-Concept Token Transformer for Medical Image Analysis","repo_url":"https://github.com/CristianoPatricio/viconex-med","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}