Researchers have developed BAC-JEPA, a novel framework for segmenting breast arterial calcifications (BAC) on mammograms using synthetic data. This label-efficient approach leverages procedurally generated arterial calcifications and their corresponding masks, trained with self-supervised Vision Transformer encoders and a convolutional decoder. The system demonstrated strong performance on synthetic validation data and achieved an AUROC of 0.8719 for image-level classification on a human-labeled dataset, indicating its potential for cardiovascular risk assessment without extensive manual annotation. AI
IMPACT Enables more efficient development of AI tools for cardiovascular risk assessment from medical imaging.
RANK_REASON The cluster describes a research paper detailing a new AI framework for medical image segmentation. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
- BAC-JEPA
- BacSeg
- Breast arterial calcifications: association with diabetes mellitus and cardiovascular mortality
- RTX 5090
- vision transformer
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
