Shape-guided Dual Consistency Semi-supervised Learning Framework for 3D Medical Image Segmentation

Usage

1. Clone the repository;

git clone https://github.com/SUST-reynole/SDC-SSL.git

2. Put the data in './data';

3. Train the model;

cd code
# e.g., for 20% labels on LA
python ./code/train_LA.py --root_path ../data/2018LA_Seg_Training_Set/ --max_iterations 6000 --labelnum 16

4. Test the model;

cd code
# e.g., for 20% labels on LA
python ./code/test_LA.py --root_path ../data/2018LA_Seg_Training_Set/ --gpu 0

Acknowledgements:

Our code is origin from UAMT, SASSNet, DTC, DTML, URPC, SSL4MIS and MC-Net. Thanks for these authors for their valuable works and hope our model can promote the relevant research as well.

Questions