Meet NeuroVFM: A New Neuroimaging Foundation Model Trained With Vol-JEPA on Uncurated Clinical MRI and CT Volumes
University of Michigan just trained a foundation model on 5.24M clinical brain scans. No radiology labels needed.
Why it matters
NeuroVFM represents a significant capability advance in medical AI—self-supervised learning on unlabeled volumetric imaging could unlock faster model deployment across radiology workflows and reduce labeling bottlenecks that plague healthcare AI adoption.
The key facts
10 to know5.24M clinical MRI and CT volumes in training dataset
Vol-JEPA architecture extends I-JEPA and V-JEPA to 3D volumetric medical imaging
Trained on uncurated clinical data without radiology-report labels
Learns brain anatomy and pathology representations
University of Michigan research origin
Self-supervised learning approach reduces annotation dependency
Vol-JEPA architecture extends I-JEPA and V-JEPA to volumetric medical imaging
Trained without radiology-report labels (unsupervised/self-supervised approach)
University of Michigan research (academic, not commercial release)
Learns brain anatomy and pathology recognition from uncurated data
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: NeuroVFM is a generalist neuroimaging foundation model from the University of Michigan, trained on 5.24M clinical MRI and CT volumes. Its Vol-JEPA base extends I-JEPA and V-JEPA to volumetric medical imaging, learning brain anatomy and pathology without radiology-report labels.