Related Experiment Video
Updated: Nov 29, 2025
10:14
Author Spotlight: Ex Vivo OCT-Based Multimodal Imaging of Human Donor Eyes for Research into Age-Related Macular Degeneration
Published on: May 26, 2023
4.0K
The SUSTech-SYSU dataset for automated exudate detection and diabetic retinopathy grading
Li Lin1,2, Meng Li3, Yijin Huang1
1Department of Electrical and Electronic Engineering, Southern University of Science and Technology, Shenzhen, 5180000, China.
Scientific Data
|November 21, 2020
Summary
This study introduces a new dataset of 1219 fundus images for diabetic retinopathy (DR) detection. The dataset aids in developing more reliable automated screening tools, overcoming limitations of current supervised methods.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Diabetic retinopathy (DR) screening relies on automated exudate detection from fundus images.
- Supervised and semi-supervised learning methods are common but require large, well-annotated datasets.
- Limitations in data size and annotation quality hinder current DR detection algorithms.
Purpose of the Study:
- To address the limitations of existing DR detection methods by creating a comprehensive annotated dataset.
- To facilitate the development and analysis of automated algorithms for DR screening and evaluation.
- To provide a valuable resource for research in exudate detection, DR grading, and landmark localization.
Main Methods:
- Construction of a dataset with 1219 fundus images from DR patients and healthy controls.
- Annotation of exudate lesions within the fundus images.
- Inclusion of additional labels: eye laterality, DR grade (3 protocols), optic disc (OD) bounding box, and fovea location.
Main Results:
- A novel, richly annotated dataset for DR analysis is presented.
- The dataset enables evaluation of various algorithms for exudate detection, OD detection, fovea localization, and DR classification.
- Facilitates research into supervised and semi-supervised learning for DR detection.
Conclusions:
- The newly created dataset is crucial for advancing automated diabetic retinopathy screening.
- It supports the development of more robust and reliable algorithms by overcoming data limitations.
- This resource will accelerate progress in computer-aided diagnosis for diabetic eye diseases.
