This is the official implementation of the paper Diffeomorphism Matching for Fast Unsupervised Pretraining on Radiographs
To train model on 4 GPUs, run
python train_dm.py path_to_csv
CheXpert pretrained model is available at https://drive.google.com/u/0/uc?export=download&confirm=C4G4&id=1-3njBZ4N9VvrfoYRIFVKt9X7fEXPmzvQ
To evaluate RegChest on CheXpert valid set, use the following commands
python evaluate.py path_to_chexpert_folder --checkpoint path_to_checkpoint
The result should be like this
| Abnormality | AUC |
|---|---|
| Enlarged Cardiomediastinum | 0.722055 |
| Cardiomegaly | 0.833717 |
| Lung Opacity | 0.926146 |
| Lung Lesion | 1.000000 |
| Edema | 0.919929 |
| Consolidation | 0.876828 |
| Pneumonia | 0.855088 |
| Atelectasis | 0.857305 |
| Pneumothorax | 0.971239 |
| Pleural Effusion | 0.932523 |
| Pleural Other | 0.995708 |
| Support Devices | 0.952462 |