Chest X-Ray Classification

Detect common thoracic diseases from chest X-ray images, using the ChestX-ray8 dataset.

XAI demos

There are some jupyter notebooks with prediction samples alongside a heatmap over the image, to explain the prediction.

CAM examples
LIME examples
IntegratedGradients and DeepLIFT examples, using the captum library.

Usage

Setup

Setup pytorch. Tested with pytorch v1.4.0 and torchvision v0.5.0.
Download the ChestX-ray8 dataset here.
Extract the images into directory <folder>/dataset/images.
Put Data_Entry_2017.csv, BBox_List_2017.csv, test_list.txt, and train_val_list.txt into directory <folder>/dataset/.

To generate train, validation, and test data entry.

python label_gen.py

This will separate train_val_list.txt into train_list.txt and val_list.txt.
3 csv files train_label.csv, val_label.csv, and test_label.csv will be generated as data entry.

Training

To train models:

    python train.py

See python train.py --help for hyper-parameter options.

References

ChestX-ray8 dataset Xiaosong Wang, Yifan Peng, Le Lu, Zhiyong Lu, Mohammadhadi Bagheri, Ronald M. Summers. ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly- Supervised Classification and Localization of Common Thorax Diseases, IEEE CVPR, pp. 3462-3471,2017
CAM TODO