DeepDiscovery is a deep learning based package focused on medical data analysis, with an emphasis on speed and robustness. We hope to provide basic capabilities to help develop tools that can be used for both scientific discovery and application in the clinic. Both a comprehensive Python API and an easy to use Qt-based GUI are planned. With the end of the Theano project, we are switching DeepDiscovery to use TensorFlow. Currently the 2D and 3D UNets and 2D UNet based segmenter have been ported. Try the corpus callosum segmentation demo in the examples! Requires Python 3.x, which can be obtained from python.org if your system doesn't already have it. pip should install the other requirements, although on some systems you may need to install TensorFlow yourself.