RANDGAN for semi supervised detection of COVID-19 in Chest X-rays

RANDGAN is a generative adversarial model implemented in Python for classification of COVID-19 Positive and COVID-19 Negative Chest X-rays.

If you use our model or the segmented COVIDx dataset, Please cite our paper https://arxiv.org/abs/2010.06418

• To generate the segmented COVIDx dataset, please refer to https://github.com/IMICSLab/Covidx-IMICS-Lung-Segmentation

* Your directory structure should be as follows:
.
├── RANDGAN_model.py
├── main.py
├── result     #generated images at each iteration are saved here
├── weight
├── modified   #direcvtory where anomaly scores are saved in
├── data                    
│   ├── COVID_test.npy              
│   ├── Normal_train.npy              
│   ├── Pneumonia_train.npy             
│   └── ...                #train and etst numpy arrays
└── ...

Train the model

• By running main.py (setting line 23 to train), model starts training

• make sure you change the file / directory paths to your local environment

Test the model

• By setting main.py to test mode (line 23), you can load in test images and calculate anonmaly scores for each image (saved to modified folder as CSV file.

Requirements

The main requirements are listed below:

Tested with Keras 2.3.1
Python 3.6
OpenCV 3.4.2
scikit-image 0.16.2
Numpy
Scikit-Learn
Matplotlib

RANDGAN Contributors

• iMICS Lab, University of Toronto, Canada https://imics.ca/
  • Saman Motamed
  • Farzad Khalvati
  • Ernest Khashayar Namdar
  • Patrik Rogalla

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.