Installation¶
FastSurfer works the same way on every system: you call run_fastsurfer.sh with a T1-weighted MRI image. How you
install it depends on your system. Pick yours:
Apple silicon: the installer package, with all software included.
Intel Macs: Docker.
Apptainer/Singularity or Docker images, for NVIDIA GPUs, AMD GPUs (experimental) or CPU only.
Docker in WSL2, for NVIDIA GPUs or CPU only.
A native installation on Ubuntu, for developers and systems without containers.
Which method should I use?¶
Your system |
Recommended |
Alternatives |
|---|---|---|
Mac with Apple silicon (M1 or newer) |
||
Mac with an Intel CPU |
||
Linux with NVIDIA GPU (workstation or cluster) |
||
Linux with AMD GPU |
||
Linux without GPU |
||
Windows |
The containers and the macOS package include all software FastSurfer needs (only the FreeSurfer license for the surface reconstruction has to be obtained separately), and the containers are what we test and validate FastSurfer with (Ubuntu 24.04). A native installation depends on the software on your system, so its results can differ from ours, and we may not be able to help if it does not work.
Before you start¶
Hardware¶
A GPU makes the segmentation much faster. How much memory FastSurfer needs depends on the voxel size and on whether the GPU or the CPU does the work; the system requirements list both.
FreeSurfer license¶
The surface pipeline uses some FreeSurfer tools, so it needs a FreeSurfer license file, as does the Talairach
registration in the segmentation (--tal_reg, used for the estimated total intracranial volume, eTIV). A
segmentation without --tal_reg does not need one.
The license is free: register at the FreeSurfer website and
you receive it by email. Save the file in your home folder and pass it to FastSurfer with
--fs_license <freesurfer_license_path>, or set the FS_LICENSE environment variable to its path. A container also
needs access to the file, see the examples on the page of your system.