Diabetic retinopathy (DR), as a debilitating ocular complication, necessitates prompt intervention and treatment. Despite the effectiveness of artificial intelligence in assisting DR grading, the interpretability of DR lesion segmentation crucial for DR grading is severely impeded by the scarcity of datasets. This paper presents and delineates TJDR, a high-quality pixel-level annotated dataset for DR. The dataset comprises 561 color fundus images sourced from the Tongji Hospital Affiliated to Tongji University. These images, captured using diverse fundus cameras including Topcon's TRC-50DX and Zeiss CLARUS 500, exhibit high resolution. For the sake of adhering strictly to principles of data privacy, the private information of images is meticulously removed while ensuring clarity in displaying anatomical structures such as the optic disc, retinal blood vessels, and macular fovea. The diabetic retinopathy lesions are annotated using the Labelme tool, encompassing four prevalent diabetic retinopathy lesions: Microaneurysms (MA), Hemorrhages (HE), Hard Exudates (EX), and Soft Exudates (SE). Meanwhile, experienced ophthalmologists conduct the annotation work with rigorous quality assurance, culminating in the construction of this dataset. This dataset has been partitioned into training and testing sets and publicly released to contribute to advancements in the diabetic retinopathy lesion segmentation research community.
If you find this dataset is useful to your research, please consider to cite our paper.
@article{mao2023tjdr,
title={TJDR: A High-Quality Diabetic Retinopathy Pixel-Level Annotation Dataset},
author={Jingxin Mao and Xiaoyu Ma and Yanlong Bi and Rongqing Zhang},
journal={arXiv preprint arXiv:2312.15389},
year={2023},
}