Skip to content

Dataset · FOMO-MRI

306K 3D Brain MRIs for SSL

FOMO-MRI/FOMO300K

306,207 3D brain MRI scans with clinical data for self-supervised learning.

What's inside

A large heterogeneous collection of 3D brain MRI scans gathered for large-scale self-supervised learning with clinical data. Version V1.1 includes 306,207 scans; the MGH-Wild subset changed its license and is no longer part of FOMO300K. Tasks include image feature extraction and zero-shot classification; accompanied by a research paper in Nature and on arXiv. Modalities are 3D and images.

What you can do with it

  • Unsupervised pretraining of 3D MRI encoders
  • Zero-shot classification of MRI scans
  • Feature extraction for clinical models
Tasks
Image feature extraction, Zero shot classification
Size
100K–1M rows
Downloads this month
6.4K
Likes
122
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("FOMO-MRI/FOMO300K")