BioMedParse
Text-prompted 3D segmentation with Microsoft BioMedParse v2.
Model card
- Input
- A 3D scan or a 2D medical image.
- Your guidance
- Name the structures you want to find in a text prompt.
- Output
- A segmentation map marking the requested structures.
- Runs on
- macOS 14.0 or later on Apple Silicon
Downloads and memory
Download size and peak memory use are not published for this version.
Before you run
The model was trained for 3D scans. It accepts 2D images, but its accuracy on them has not been established.
License and terms
Software: Apache-2.0. Model files have their own license terms. BioMedParse models are for noncommercial use under CC-BY-NC-SA-4.0.
More segmentation
Intended use
For research and education. Not for clinical diagnosis or treatment decisions. Outputs require human review.