Utilities and tools for machine learning on BOLD-fMRI data
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
.github
docs
img
skbold
.gitignore
.travis.yml
CHANGELOG.rst
LICENSE
MANIFEST.in
README.rst
requirements.txt
run_pytest.sh
setup.cfg
setup.py

README.rst

skbold - utilities and tools for machine learning on BOLD-fMRI data

https://travis-ci.org/lukassnoek/skbold.svg?branch=master Documentation Status https://coveralls.io/repos/github/lukassnoek/skbold/badge.svg?branch=develop

The Python package skbold offers a set of tools and utilities for machine learning analyses of functional MRI (BOLD-fMRI) data. Instead of (largely) reinventing the wheel, this package builds upon an existing machine learning framework in Python: scikit-learn. The modules of skbold are applicable in several 'stages' of typical pattern analyses (see image below), including pattern estimation, data representation, pattern preprocessing, feature selection/extraction, and model evaluation/feature visualization.

img/scope.png

Documentation

Please see skbold's ReadTheDocs page for more info on how to use skbold!

Installation & dependencies

Although the package is very much in development, it can be installed using pip:

$ pip install skbold

However, the pip-version is likely behind compared to the code on Github, so to get the most up to date version, use git:

$ pip install git+https://github.com/lukassnoek/skbold.git@master

Skbold is largely Python-only (both Python2.7 and Python3) and is built around the "PyData" stack, including:

  • Numpy
  • Scipy
  • Pandas
  • Scikit-learn

And it uses the awesome nibabel package for reading/writing nifti-files. Also, skbold uses FSL (primarily the FLIRT and applywarp functions) to transform files from functional (native) to standard (here: MNI152 2mm) space. These FSL-calls are embedded in the convert2epi and convert2mni functions, so avoid this functionality if you don't have a working FSL installation.

Authors & credits

This package is being develop by Lukas Snoek from the University of Amsterdam with contributions from Steven Miletic and help from Joost van Amersfoort.

License and citing skbold

The code is BSD (3-clause) licensed. If you use 'skbold' in your research, I would appreciate if you'd reference the following:

Snoek, L. (2017). Skbold: utilities and tools for machine learning on BOLD-fMRI data (version 0.4.0). https://doi.org/10.5281/zenodo.1064090.

Suggestions, issues, and contributing

If you have suggestions or issues, please submit an issue to the issue-tracker. Also, I would love contributions to skbold from others! In case you want to contribute, fork this repository and submit a pull-request with your contribution!