BrainNext foundation model achieves top ranks in brain MRI analysis challenge

Researchers have developed BrainNext, a self-supervised foundation model designed for analyzing brain MRI scans. This model utilizes a three-dimensional Bi-Directional xLSTM-UNet architecture and was pretrained on over 60,000 unlabeled brain MRI examinations. BrainNext demonstrated strong performance in the FOMO 2025 challenge, achieving second place overall and first place in meningioma segmentation, highlighting its effectiveness in transferring knowledge to diverse neuroimaging tasks. AI

IMPACT This model's success in the FOMO 2025 challenge suggests a significant advancement in self-supervised learning for medical imaging, potentially accelerating research and clinical applications in brain MRI analysis.

RANK_REASON The cluster describes a new research paper detailing a novel foundation model for medical imaging analysis, including its architecture, training data, and performance on a specific challenge. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Bi-Directional xLSTM-UNet
• BrainNext
• CORE Recommender
• DagsHub
• FOMO 2025
• Hugging Face
• Moona Mazher Mazher

• paper
• model release
• product

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