SCL Training & validation

Prepare

Step 1. prepare for data

You need to process the WSI into the following format. The processing method can be found in https://github.com/mahmoodlab/CLAM

The CTransPath feature extractor and the pretrained model can be download in

DATA_DIR
├─patch_coord
│      slide_id_1.h5
│      slide_id_2.h5
│      ...
└─patch_feature
        slide_id_1.pt
        slide_id_2.pt
        ...

The h5 file in the patch_coord folder contains the coordinates of each patch of the WSI, which can be read as

coords = h5py.File(coords_path, 'r')['coords'][:]
# coords is a array like:
# [[x1, y1], [x2, y2], ...]

The pt file in the patch_featurefolder contains the features of each patch of the WSI, which can be read as

features = torch.load(features_path, map_location=torch.device('cpu'))
# features is a tensor with dimension N*F, and if features are extracted using CTransPath, F is 768

Step 2. preparing the data set split

You need to divide the dataset into a training set validation set and a test set, and store them in the following format

SPLIT_DIR
    test_set.csv
    train_set.csv
    val_set.csv

And, the format of the csv file is as follows

slide_id	label
slide_id_1	0
slide_id_2	1
...	...

Train model

Step 1. create a config file

We have prepared two config file templates (see ./configs/) for SCL-WC, like

General:
    seed: 7
    work_dir: WORK_DIR
    fold_num: 4

Data:
    split_dir: SPLIT_DIR
    data_dir_1: DATA_DIR_1 
    features_size: 768
    n_classes: 2

Model:
    network: 'SCL'

In the config, the correspondence between the Model.network, Train.training_method and Train.val_method is as follows

Model.network	Train.training_method	Train.val_method
SCL	SCL	SCL

Step 2. train model

Run the following command

python train.py --config_path [config path] --begin [begin index] --end [end index]

--begin and --end used to control repetitive experiments

License

SCL is released under the GPLv3 License and is available for non-commercial academic purposes.

Citation

Please use below to cite this paper if you find our work useful in your research.

@article{wang2022scl,
  title={Scl-wc: Cross-slide contrastive learning for weakly-supervised whole-slide image classification},
  author={Wang, Xiyue and Xiang, Jinxi and Zhang, Jun and Yang, Sen and Yang, Zhongyi and Wang, Ming-Hui and Zhang, Jing and Yang, Wei and Huang, Junzhou and Han, Xiao},
  journal={Advances in neural information processing systems},
  volume={35},
  pages={18009--18021},
  year={2022}
}