We propose a new benchmark dataset, namely Diabetic Retinopathy Two-field image Dataset (DRTiD), consisting of 3,100 two-field fundus images. Two-field images contains a pair of macula-centric and optic disc-centric images. We provide annotations of DR severity grades and localization of macula & optic disc.
Junlin Hou, Jilan Xu, Fan Xiao, Rui-Wei Zhao, Yuejie Zhang, Haidong Zou, Lina Lu, Wenwen Xue, Rui Feng. Cross-Field Transformer for Diabetic Retinopathy Grading on Two-field Fundus Images. 2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE Computer Society, 2022: 985-990. [paper][arxiv]
Source code of CrossFiT: https://github.com/FDU-VTS/DRTiD/tree/main/CrossFiT
Fill in the information in the application form wenjuanxing (问卷星) or GoogleForm and the download link will be feedbacked after accurate filling.
DRTiD
├── Original Images
└── Ground Truths
├── DR_grade
│ ├── a. DR_grade_Training.csv
│ └── b. DR_grade_Testing.csv
└── Optic_Macula_Localization
└──op_ma_localization.csv
All images from the DRTiD dataset are of gradable quality and annotated by three experienced ophthalmologists.
We also provide the initial version, without image quality check and label re-verification by ophthalmologists. The initial labels are provided by community screening.
An DR open access dataset for research only.
By using the DRTiD dataset, you are obliged to reference the following paper:
@inproceedings{hou2022cross,
title={Cross-Field Transformer for Diabetic Retinopathy Grading on Two-field Fundus Images},
author={Hou, Junlin and Xu, Jilan and Xiao, Fan and Zhao, Rui-Wei and Zhang, Yuejie and Zou, Haidong and Lu, Lina and Xue, Wenwen and Feng, Rui},
booktitle={2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)},
pages={985--990},
year={2022},
organization={IEEE Computer Society}
}
If you have any questions, please feel free to contact Dr. Junlin Hou (jlhou18@fudan.edu.cn).