John Garrett Dataverse provides a curated dataset of 3D MRI studies designed for evaluating foundation model embeddings. The dataset links patient demographics, study acquisition metadata, and series-level imaging parameters to precomputed 3D FM embeddings. It was last updated on May 5, 2026.
Use Cases
- Evaluate foundation model embedding performance based on linked patient demographics and imaging parameters
- Validate model generalization across clinical variation based on included patient and study data
- Train lightweight adapters for downstream tasks using precomputed 3D FM embeddings
- Analyze model strengths and weaknesses systematically across different imaging conditions
Strengths
- Dataset is curated specifically for foundation model evaluation in clinical imaging contexts
- Links multiple data types: patient demographics, study acquisition metadata, imaging parameters, and precomputed embeddings
- Supports assessment of model generalization across clinical variation
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Row count and file size are unknown, which may limit suitability assessment
Provenance
- Source
- John Garrett Dataverse
- Collection Method
- Curated dataset
- Freshness
- Last updated 2026-05-05 03:43:41