Deep learning processing framework for Keras. Partly based on DeepVoxNet.
This package can be added to your python 3.9 environment via:
- First cloning/downloading the repository and then via:
pip install -e /path/to/deepvoxnet2
- Installing it directly from Github via:
pip install git+https://github.com/JeroenBertels/deepvoxnet2
To upgrade your installation using the first method just download the latest version and repeat the process or git pull the new version. When using the second method simply repeat the command but add the --upgrade flag. You can also install/revert to a specific version; in that case append @version_tag (e.g. @deepvoxnet-2.10.23).
Some functions require the SimpleITK and SimpleElastix software to be installed. To install these packages also, please append the paths in the above commands with [sitk].
A Jupyter Notebook-style tutorial can be found here, which guides you through some of the basic design ideas behind deepvoxnet2.
Other real-world examples are:
- A notebook with all experiments and code accompanying this article about the effect of
$\Phi$ and$\epsilon$ when using the Dice loss in tasks with missing or empty labels.
Jeroen Bertels is part of NEXIS, a project that has received funding from the European Union's Horizon 2020 Research and Innovation Programme.