Researchers have developed MIRAGE, a novel 2D U-Net model designed to infer contrast enhancement in breast MRIs from pre-contrast slices. The model integrates global reconstruction and perceptual losses with specialized lesion-aware supervision, including penalties for missed tumor enhancement and multi-scale auxiliary tumor segmentation. Evaluated on 301 cases from the MAMA-SYNTH dataset, MIRAGE demonstrated superior performance across multiple metrics, significantly improving downstream lesion localization compared to existing baselines like Pix2Pix and conditional diffusion models. AI
IMPACT This research could lead to more accurate and efficient diagnosis of breast cancer through improved MRI analysis.
RANK_REASON The cluster describes a new research paper detailing a novel AI model for medical image analysis.
Read on Hugging Face Daily Papers →
- Conditional diffusion processes in population genetics
- latent bridge-matching
- lpips
- MAMA-SYNTH
- MIRAGE
- nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation
- Pix2Pix
- U-Net
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
