iChallenge-DeepDRiD-Task1
糖尿病视网膜病变（DR）是糖尿病最常见的并发症之一。DeepDRiD数据集旨在解决设计自动DR筛查系统时遇到的实际问题，推动该领域的技术进步。该数据集包含三个子挑战赛：（1）常规眼底DR分级，针对不同质量的图像；（2）图像可用性质量评估；（3）超广域（UWF）DR分级，针对不同设备拍摄的图像。本文聚焦于子挑战赛1。数据集遵循了相关程序。
yanjing可视化图片
iChallenge-DeepDRiD-Task1_0.png
数据集元信息
| 维度 | 2D |
| 模态 | fundus |
| 任务类型 | classification |
| 解剖结构 | 眼部 |
| 解剖区域 | 眼 |
| 类别数 | 5 |
| 数据量 | - |
| 文件格式 | JPG |
文件结构
.
├── regular_fundus_images
│ ├── Online-Challenge1&2-Evaluation
│ │ ├── Challenge1_upload.csv
│ │ ├── Challenge2_upload.csv
│ │ ├── Challenge1_labels.xlsx
│ │ ├── Challenge1_labels.xlsx
│ │ ├── Images
│ │ └── Readme.docx
│ ├── regular-fundus-training
│ │ ├── Images
│ │ ├── Readme.docx
│ │ └── regular-fundus-training.csv
│ └── regular-fundus-validation
│ ├── Images
│ ├── Readme.docx
│ └── regular-fundus-validation.csv
└── ultra-widefield_images
├── Online-Challenge3-Evaluation
│ ├── Challenge3_upload.csv
│ ├── Challenge3_labels.xlsx
│ ├── Images
│ └── Readme.docx
├── ultra-widefield-training
│ ├── Images
│ ├── Readme.txt
│ └── ultra-widefield-training.csv
└── ultra-widefield-validation
├── Images
├── Readme.txt
└── ultra-widefield-validation.csv图像尺寸统计
| 统计类型 | 间距 (mm) | 尺寸 |
|---|---|---|
| 最小值 | - |
(1725, 1592, 3) |
| 中位值 | - |
(1878, 1827, 3) |
| 最大值 | - |
(2232, 1592, 3) |
引用
@misc{liu2022deepdrid,
title={DeepDRiD: diabetic retinopathy-grading and image quality estimation challenge. Patterns 3 (6), 100512 (2022)},
author={Liu, R and Wang, X and Wu, Q and Dai, L and Fang, X and Yan, T and Son, J and Tang, S and Li, J and Gao, Z and others},
year={2022}
}