Skull-stripping for brain images of any type
10K+
SynthStrip is a skull-stripping tool designed to extract the brain from various medical imaging modalities, spanning contrasts, resolutions, and diverse subject populations. It leverages a deep-learning strategy that synthesizes extremely variable training images from segmentation maps, enabling the model to generalize across acquisition protocols.
For details on SynthStrip and how to use the Docker image, visit synthstrip.io.
synthstrip:1.8, latest
synthstrip:1.8-gpu
- add procps to the Docker images
- revert normalization to avoid torch.quantile error
- control and report background-fill value
synthstrip:1.7, latest
synthstrip:1.7-gpu
- introduce separate CPU-only and GPU images
- switch from Ubuntu to minimal Python base image
- add option to control the number of CPU threads
- set up automated tests
synthstrip:1.6
- add support for large borders
- resolve NumPy deprecation warning
- add option to save the distance transform
synthstrip:1.5
- support multi-frame images
- save output files as regular user with Docker
- avoid waiting for PyTorch only to show help text
synthstrip:1.4
- fix floating-point error in output voxel size
- switch from CentOS to Ubuntu base image
synthstrip:1.3
- fix output images appearing shifted in ITK-SNAP
synthstrip:1.2
- fix qform/sform handling in surfa
- add model for predicting brain masks without CSF
synthstrip:1.1
- resolve SciPy deprecation warning
- switch from freesurfer to surfa package
- print help text when no arguments provided
synthstrip:1.0
- release initial image
Reach out to the FreeSurfer mailing list or open an issue on GitHub.
Content type
Image
Digest
sha256:ebbc17722…
Size
370.9 MB
Last updated
about 1 year ago
docker pull freesurfer/synthstripPulls:
208
Last week