kbronik2017 / Multi_Label_Segmentation_UCL

Multi-Label Multi/Single-Class Image Segmentation

Multi-Label Multi/Single-Class Image Segmentation

Publication

Le Zhang, Ryutaro Tanno, Kevin Bronik, Chen Jin, Parashkev Nachev, Frederik Barkhof, Olga Ciccarelli, and Daniel C. Alexander, Learning to Segment When Experts Disagree, International Conference on Medical image computing and Computer-Assisted Intervention (MICCAI). Springer, Cham, 2020.

Click here to see full pdf file: Link to PDF

Running the GUI Program!

First, user needs to install Anaconda https://www.anaconda.com/

Then

  - conda env create -f conda_environment_Training_Inference.yml  

and

  - conda activate traintestenv  

finally

  - python  Training_Inference_GUI.py 

After lunching the graphical user interface, user will need to provide necessary information to start training/testing as follows:

Running the Program from the command line!

First

  - conda activate traintestenv  

then for training

  - python  segmentation_network_Training_without_GUI.py  [or annotation_network_Training_without_GUI.py]

for testing

  - python  segmentation_network_Inference_without_GUI.py  [or annotation_network_Inference_without_GUI.py]

Testing the Program!

Examples of Training and Testing subjects can be found in: https://github.com/UCLBrain/MSLS/tree/master/examples (which will allow users to quickly and easily train and test the program)

..................................................................................................................................................................