MICCAI 2023 UWAT-GAN

This code is the pytorch implementation of our paper "UWAT-GAN: Fundus Fluorescein Angiography Synthesis via Ultra-wide-angle Transformation Multi-scale GAN". It can be used to turning UWF scanning laser ophthalmoscopy(UWF-SLO) to the UWF fluorescein angiography(UWF-FFA) and display the tiny vascular lesion areas.

New Version

You can find our improved vision of UWAT-GAN, called UWAT-GAN-R

UWF-SLO to UWF-FA at 3k resolution

Pre-requisties

Linux
python>=3.7
NVIDIA GPU (memory>=10G) + CUDA cuDNN

Getting start to evaluate

Install dependencies

pip install -r requirements.txt

Configure the checkpoints

Fistly download the checkpoint named as the 'UWFA-GAN_checkpoints', move it into the project directory and rename it to the 'checkpoints'.

├── checkpoints
    ├──d_model_1_fine.pt
    ├──d_model_2_coarse
    ├──g_model_coarse
    ├──g_model_fine

Evaluation dataset

Due the privacy of our dataset, we only provide 4 pictures for the result viewing. They are located at './example_pics/'

├── example_pics
    ├──1.png
    ├──1-1.png
    ├──2.png
    ├──2-2.png
    ...
    ├──4-4.png

1.png means first UWF-SLO and the 1-1.png means first UWF-FA, 2 means the second pair, 3, 4, respectively.

Evaluation

To do the evaluation process, run the following command:

python inference.py

After the evaluation, some new directories may be created. the running results are saved in the directories './result_save' and two sub-directories called './Coarse_result' and './Fine_result'.

├── result_save
    ├──Coarse_result
    ├──Fine_result

The './Coarse_result' saves the results coarse generator generates, while the './Fine_result' corresponds to results fine generator generates.

Citation

@InProceedings{fang2023uwat,
    author    = {Fang, Zhaojie and Chen, Zhanghao and Wei, Pengxue and Li, Wangting and Zhang, Shaochong and Elazab, Ahmedand Jia, Gangyong and Ge, Ruiquan and Wang, Changmiao},
    title     = {UWAT-GAN: Fundus Fluorescein Angiography Synthesis via Ultra-Wide-Angle Transformation Multi-scale GAN},
    booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2023},
    month     = {October},
    year      = {2023},
    url       = {https://link.springer.com/chapter/10.1007/978-3-031-43990-2_70}
}

Benny0323/UWAT-GAN