Identifying clinically relevant findings in breast cancer using deep learning and feature attribution on local views from high-resolution mammography
These results demonstrate that CorRELAX generates interpretable coarse-segmentation maps that enhance automated lesion detection in mammography. The improved interpretability facilitates clinically reliable decision-making and addresses a critical barrier toward the integration of AI-based methods in breast cancer screening workflows.
Authors
- Leondry Mayeta-Revilla (ORCID: https://orcid.org/0009-0007-7511-2305)
- Fabián Pardo (ORCID: https://orcid.org/0000-0002-6994-0418)
- Diego Mellado (ORCID: https://orcid.org/0000-0001-8078-253X)
- Julio A Sotelo (ORCID: https://orcid.org/0000-0002-0915-5215)
- Stéren Chabert (ORCID: https://orcid.org/0000-0002-2890-5077)
- Marvin Querales (ORCID: https://orcid.org/0000-0003-2372-6050)
- Rodrigo Salas (ORCID: https://orcid.org/0000-0002-0350-6811)
- Scarlett Lever (ORCID: https://orcid.org/0009-0001-3161-3980)
- Eduardo Godoy (ORCID: https://orcid.org/0009-0009-6775-7941)
Institutions
- Viña del Mar University (CL)
- Universidad del Mar (CL)
- Pontificial Catholic University of Valparaiso (CL)
- Universidad de Aconcagua (CL)
- Millennium Institute for Integrative Biology (CL)
- Federico Santa María Technical University (CL)
- University of Valparaíso (CL)
Publication Details
- Journal
- Frontiers in Oncology
- Published
- 2025-09-23
- DOI
- https://doi.org/10.3389/fonc.2025.1601929
- Citations
- 1
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 2.48