arXiv:2603.14644v3 Announce Type: replace-cross Abstract: Publicly available full-field digital mammography (FFDM) datasets remain limited in size, clinical annotations, and vendor diversity, hindering the development of robust models. We introduce LUMINA, a curated, multi-vendor FFDM dataset that explicitly encodes acquisition energy and vendor metadata to capture clinically relevant appearance variations often overlooked in existing benchmarks. This dataset contains 1824 images from 468 patients (960 benign, 864 malignant), with pathology-confirmed labels, BI-RADS assessments, and breast-den
The proliferation of AI in medical imaging necessitates more robust, diverse, and representative datasets to overcome limitations of existing benchmarks.
The introduction of LUMINA addresses a critical gap in medical AI development by providing a multi-vendor, energy-harmonized mammography benchmark, improving model generalizability and clinical relevance.
AI models for mammography can now be trained and evaluated on a more diverse and clinically representative dataset, leading to improved diagnostic accuracy across different equipment and patient populations.
- · Medical AI developers
- · Diagnostic imaging companies
- · Patients needing early cancer detection
- · Developers relying on limited, proprietary datasets
- · Existing less diverse mammography benchmarks
Improved performance and broader applicability of AI-powered mammography diagnostics.
Accelerated adoption of AI in clinical settings due to increased trust in model robustness and reduced vendor lock-in.
Potential for earlier and more accurate breast cancer detection globally, leading to better patient outcomes and reduced healthcare costs.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG