Amarachi B. Mbakwe,Lyuyang Wang, Mehdi Moradi, and Ismini Lourentzou

Here, we provide the code implementation of the paper: Hierarchical Vision Transformers for CXR Disease Progression Detection.

Python 3.8.0 pytorch 1.10.1 torchvision 0.11.2 einops 0.3.2 

To run the code, create a virtual conda environment named CheXRelFormer with the following cmd:

conda create --name CheXRelFormer --file requirements.txt conda activate CheXRelFormer 

Clone this repo:

git clone https://github.com/PLAN-Lab/CheXRelFormer.git cd CheXRelFormer

To train the models, edit the arguments in the run_CheXRelFormer.sh in the script folder. Then run the training script by running the command sh scripts/run_CheXRelFormer.sh.

To evaluate the models, edit the arguments in the eval_CheXRelFormer.sh in the script folder. Then run the training script by running the command sh scripts/eval_CheXRelFormer.sh.

We used the following dataset:

• Chest Imagenome
• MIMIC-CXR-JPG

""" The dataset folder was processed in the following structure； ├─A ├─B ├─label └─list """ 

A: previous CXR from a patient;

B:post images CXR from the same patient;

label: comparison - improved, worsened, no change;

list: contains train.txt, val.txt and test.txt, each file contains the image names.

If you find this method and/or code useful, please consider citing

@inproceedings{10.1007/978-3-031-43904-9_66, author="Mbakwe, Amarachi B. and Wang, Lyuyang and Moradi, Mehdi and Lourentzou, Ismini", title="Hierarchical Vision Transformers for Disease Progression Detection in Chest X-Ray Images", booktitle="Medical Image Computing and Computer Assisted Intervention -- MICCAI 2023", year="2023", publisher="Springer Nature Switzerland", address="Cham", pages="685--695" }