HookNet

Multi-resolution convolutional neural networks for semantic segmentation in histopathology whole-slide images.

#ExaMode

Training

dependecies

• This code has been tested on Ubuntu 18.04, keras==2.0.8 and tensorflow-gpu==1.14

Examples

• train.py will train HookNet on random values. Please adjust the script with your own batchgenerator or sampling function.
• For an explanation about possible settings see the comments in parameters.yml. All settings defined in parameters.yml can be overwritten via command line arguments (see argconfigparser.py for more info).

Inference

dependecies

• Inference depends on the python api multiresolutionimageinterface.py from ASAP (https://github.com/computationalpathologygroup/ASAP/releases).

Examples

• apply.py in this repository will apply a trained hooknet on a WSI.

Additional Information

For more information, please check the code comments and the doc strings. If you happen to experience any problems, have questions, or would like to give feedback, feel free to open an issue.

A pretraind model on breast or colon can be applied via https://grand-challenge.org/. Please create an user account and request access to an algorith if you are interested.

You can try out a pretrained HookNet on breast tissue here:
https://grand-challenge.org/algorithms/hooknet/

You can try out a pretrained HookNet on colon tissue here:
https://grand-challenge.org/algorithms/hooknet-colon/