Datasets
Standard Dataset
High-Resolution Retinal Fundus Dataset for Diabetic Retinopathy Detection from Aizawl
- Citation Author(s):
- Submitted by:
- C Vanlalnunpuia
- Last updated:
- Mon, 01/20/2025 - 15:00
- DOI:
- 10.21227/c20k-b753
- Data Format:
- License:
62 Views- Categories:
- Keywords:
Abstract
This dataset contains high-resolution retinal fundus images collected from 495 unique subjects from Eye Care hospital in Aizawl, Mizoram, for diabetic retinopathy (DR) detection and classification. The images were captured over five years using the OCT RS 330 device, which features a 45° field of view (33° for small-pupil imaging), a focal length of 45.7 mm, and a 6.25 mm sensor width. Each image was acquired at a resolution of 3000x3000 pixels, ensuring high diagnostic quality and the visibility of subtle features like microaneurysms, exudates, and hemorrhages. Annotations were conducted by an experienced ophthalmologist using established DR grading standards. Each image is categorized into one of five (0-4) DR severity levels: no DR, mild DR, moderate DR, severe DR, and proliferative DR. The dataset reflects the unique demographic and healthcare characteristics of Mizoram, addressing the gap in representation of underrepresented populations in existing DR research. This dataset provides an invaluable resource for advancing AI-based DR detection, with applications in classification, segmentation, and augmentation studies.
Dataset Structure
The dataset is organized into a single folder DR_images containing the images and a .CSV file containing the labels.
Metadata File
The dataset includes a metadata file (dr_labels.csv) with the following columns:
id_code: Unique identifier for each image file.
Label: The severity level of diabetic retinopathy (0–4).
File Format
Images: JPG format, resolution 3000x3000 pixels.
Size: Ranging from 465kb - 872kb
Metadata: CSV format for easy integration into data pipelines.
Usage Steps
Download the dataset from the provided IEEE Dataport link.
Unzip the files into a local directory.
Refer to the metadata.csv file for labels and additional information.
Use the dataset for tasks such as:
Training and evaluating classification models.
Image preprocessing and segmentation studies.
Data augmentation and imbalance mitigation research.
Contact Information:
For queries or additional information, please contact [cvanlalnunpuia8@gmail.com, puitearalte@hatim.ac.in].
Comments
This dataset supports the research described in ["Advancing Diabetic Retinopathy Research with a Dataset from Aizawl region of Mizoram", to be published]