New LUMINA mammography benchmark dataset released with harmonization protocol

Researchers have introduced LUMINA, a new benchmark dataset for mammography AI that addresses limitations in existing datasets by including diverse vendors and acquisition energies. The dataset comprises 1824 images from 468 patients, with detailed annotations including pathology, BI-RADS assessments, and breast density. To handle variations, a novel energy harmonization method is proposed and benchmarked against CNN and transformer models, showing improved performance and more localized diagnostic insights. AI

IMPACT This new benchmark and harmonization protocol could lead to more robust and deployable AI models for mammography, improving diagnostic accuracy and consistency across different imaging systems.

RANK_REASON The cluster describes a new academic paper introducing a novel dataset and methodology for AI in medical imaging. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• CNN
• EfficientNet-B0
• Hongyi Pan
• LUMINA
• transformer

• paper
• product
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →