MRI Brain and Body Segmentations from the MR-RATE Foundation Model

MR-RATE-nvseg-ctmr provides native-space voxel-wise multi-label segmentations for brain and body structures from MRI volumes. The dataset was created by Forithmus and released in April 2026. It is part of the larger MR-RATE vision-language foundation model and dataset release for magnetic resonance imaging.

Use Cases

• Train or validate brain segmentation models using native-space voxel-wise multi-label predictions for center modality MRI volumes.
• Develop body segmentation algorithms using predicted segmentations for all MRI volumes in the collection.
• Benchmark vision-language models for radiology by linking segmentation masks with associated radiology reports from the full MR-RATE dataset.
• Analyze anatomical structure correlations by examining multi-label segmentation data across different MRI volumes.

Strengths

• Part of a comprehensive vision-language foundation model release for MRI, indicating a structured research effort.
• Provides both brain and body segmentations, covering multiple anatomical regions of interest.

Limitations

• Specific dataset size, row count, and file formats are not detailed in the provided description.
• The sample data and specific column structures are unavailable, complicating initial assessment.

Provenance

Source

Forithmus on Hugging Face.

Collection Method

Segmentations are predicted outputs, likely generated by an AI model for MRI volumes.

Freshness

Last updated on April 4, 2026, indicating recent maintenance.