PREDICT 2.1.0

Predict: a Radiomics Extensive D.... Interchangable Classification Toolkit.

This is an open-source python package supporting Radiomics medical image feature extraction and classification.

We aim to add a wide variety of features and classifiers to address a wide variety classification problems. Through a modular setup, these can easily be interchanged and compared.

Documentation

For more information, see the sphinx generated documentation available here (WIP).

Alternatively, you can generate the documentation by checking out the master branch and running from the root directory:

python setup.py build_sphinx

The documentation can then be viewed in a browser by opening PACKAGE_ROOT\build\sphinx\html\index.html.

Installation

PREDICT has currently only been tested on Unix with Python 2.7. The package can be installed through the setup file:

python setup.py install

Make sure you first install the required packages:

pip install -r requirements.txts

FASTR tools

When running the FASTR package with a version lower than 1.3.0, you need to manually add the PREDICT fastr_tools path to the FASTR tools path. Go the your FASTR config file (default: ~/.fastr/config.py) and add the fastr_tools path analogue to the description in the PREDICT/fastrconfig/PREDICT_config.py file:

packagedir = site.getsitepackages()[0]
tools_path = [os.path.join(packagedir, 'PREDICT', 'fastr_tools')] + tools_path

When using FASTR >1.3.0, the PREDICT config file will be automatically created for you in the default: ~/.fastr/config.d folder.

Note that the Python site package does not work properly in virtual environments. You must then manually locate the packagedir.

Preprocessing

From version 1.0.2 and on, preprocessing has been removed from PREDICT. It is now available as a separate tool in the WORC package, as it’s also a separate step in the radiomics workflow. We do advice to use the preprocessing function and thus also WORC.

3rd-party packages used in PREDICT:

We mainly rely on the following packages:

• SimpleITK (Image loading and preprocessing)

• numpy (Feature computation)

• sklearn, scipy (Classification)

• FASTR (Fast and parallel workflow execution)

• pandas (Storage)

• PyRadiomics

See also the requirements file.

License

This package is covered by the open source APACHE 2.0 License.

Contact

We are happy to help you with any questions: please send us a message or create an issue on Github.