BrainChop is a lightweight tool for brain segmentation that runs on pretty much everything.
You can install BrainChop using pip:
pip install brainchop
To use BrainChop, run the following command:
brainchop input.nii.gz -o output.nii.gz
You can also install BrainChop using docker
git clone git@github.com:neuroneural/brainchop-cli.git
cd brainchop-cli
docker build -t brainchop .
Then to run, use
docker run --rm -it --device=nvidia.com/gpu=all -v [[output directory]]:/app brainchop [[input nifti file]] -o [[output nifti file]]
On some systems (like recent 25.05 nixos), the docker run command will need to be prepended with
docker run --rm -it --device=nvidia.com/gpu=all
Where:
input.nii.gz
is your input NIfTI fileoutput.nii.gz
is the desired output file name
- Python 3.6+
- tinygrad : our tiny and portable (but powerful) ML inference engine
- numpy : basic tensor operations
- nibabel : to read nifti files
- requests : to download models
This project is licensed under the MIT License.