Our BrainPrint tools provide shape descriptors of neuroanatomical structures and require a FreeSurfer or FastSurfer segmentation as a pre-processing step. BrainPrint is based on “ShapeDNA” a spectral shape descriptor.
This is the brainprint python package, a derivative of the original
BrainPrint-legacy scripts,
with the primary goal to provide a Python-only version, to integrate the
LaPy package, and to remove dependencies
on third-party software (shapeDNA-* binaries, gmsh, meshfix). As a result,
some functionality of the original BrainPrint-legacy scripts is no longer
maintained (currently no support of tetrahedral meshes and no support of
cortical parcellations or label files).
Use the following code to install the latest release of LaPy into your local Python package directory:
python3 -m pip install brainprint
This will also install the necessary dependencies, e.g. the LaPy package. You may need to add your local Python package directory to your $PATH in order to run the scripts.
Once installed, the package provides a brainprint executable which can be run from the command line.
The brainprint CLI enables per-subject computation of the individual brainprint descriptors. Its usage and options are summarized below;
detailed info is available by calling the script without any arguments from the command line.
brainprint --sdir <directory> --sid <SubjectID> [--num <num>] [--evec] [--skipcortex] [--norm <surface|volume|geometry|none> ] [--reweight] [--asymmetry] [--outdir <directory>] [--help] [--more-help]
Options:
--help Show this help message and exit
--more-help Show extensive help message and exit
Required options:
--sid <SubjectID>
Subject ID (FreeSurfer-processed directory inside the
subjects directory)
--sdir <directory>
FreeSurfer subjects directory
Processing directives:
--num <num> Number of eigenvalues/vectors to compute (default: 50)
--evec Switch on eigenvector computation (default: off)
--skipcortex Skip cortical surfaces (default: off)
--norm <surface|volume|geometry|none>
Switch on eigenvalue normalization; will be either surface,
volume, or determined by the geometry of the object. Use
"none" or leave out entirely to skip normalization.
--reweight Switch on eigenvalue reweighting (default: off)
--asymmetry Perform left-right asymmetry calculation (default: off)
--cholmod Switch on use of (faster) Cholesky decomposition instead
of (slower) LU decomposition (default: off). May require
manual install of scikit-sparse package.
Output parameters:
--outdir=OUTDIR Output directory (default: <sdir>/<sid>/brainprint)
--keep-temp Whether to keep the temporary files directory or not
by default False
brainprint can also be run within a pure Python environment, i.e. installed and imported as a Python package. E.g.:
>>> from brainprint import Brainprint
>>> subjects_dir = "/path/to/freesurfer/subjects_dir/"
>>> subject_id = "42"
>>> bp = Brainprint(subjects_dir=subjects_dir, asymmetry=True, keep_eigenvectors=True)
>>> results = bp.run(subject_id=subject_id)
>>> results
{"eigenvalues": PosixPath("/path/to/freesurfer/subjects_dir/subject_id/brainprint/subject_id.brainprint.csv"), "eigenvectors": PosixPath("/path/to/freesurfer/subjects_dir/subject_id/brainprint/eigenvectors"), "distances": PosixPath("/path/to/freesurfer/subjects_dir/subject_id/brainprint/subject_id.brainprint.asymmetry.csv")}
The script will create an output directory that contains a CSV table with
values (in that order) for the area, volume, and first n eigenvalues per each
FreeSurfer structure. An additional output file will be created if the
asymmetry calculation is performed and/or for the eigenvectors (CLI --evecs flag or keep_eigenvectors on class initialization).
There are some changes in functionality in comparison to the original BrainPrint scripts:
The API Documentation can be found at https://deep-mi.org/BrainPrint .
If you use this software for a publication please cite:
[1] BrainPrint: a discriminative characterization of brain morphology. Wachinger C, Golland P, Kremen W, Fischl B, Reuter M. Neuroimage. 2015;109:232-48. http://dx.doi.org/10.1016/j.neuroimage.2015.01.032 http://www.ncbi.nlm.nih.gov/pubmed/25613439
[2] Laplace-Beltrami spectra as 'Shape-DNA' of surfaces and solids. Reuter M, Wolter F-E, Peinecke N Computer-Aided Design. 2006;38:342-366. http://dx.doi.org/10.1016/j.cad.2005.10.011
FastSurfer is a fast and accurate deep-learning pipeline for the analysis of human brain MRI. FastSurfer provides a fully compatible FreeSurfer alternative for volumetric and surface-based thickness analysis, also supporting sub-mm resolutions, and sub-segmentation of neuroanatomical structures.
LaPy is an open-source Python package for differential geometry on triangle and tetrahedra meshes. It includes an FEM solver to estimate the Laplace, Poisson or Heat equations, also the computations of gradients, divergence, mean-curvature flow, conformal mappings, geodesics, ShapeDNA and more.