50,000 brain MRI scans form a large-scale dataset for self-supervised learning in medical imaging. The dataset, created by FOMO-MRI, was released as part of the FOMO25 Foundation Model Challenge at MICCAI 2025. A preprint paper describing the dataset was published in 2025.
Use Cases
- Training self-supervised vision models based on the large collection of 3D brain MRI scans.
- Developing and benchmarking foundation models for brain MRI analysis as referenced in the MICCAI 2025 challenge.
- Conducting research on heterogeneous clinical data integration using the associated metadata implied by the description.
Strengths
- Contains 50,000 brain MRI scans, providing substantial scale for model training.
- Designed for and used in a major academic challenge (FOMO25 at MICCAI 2025), indicating community relevance.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- FOMO-MRI
- Collection Method
- Likely aggregated from clinical sources, as indicated by the mention of heterogeneous data.
- Time Range
- null
- Freshness
- Last updated 2026-04-10 05:39:27; freshness should be verified.
- Geography
- null