Conjunctival and Retinal Images of Healthy Subjects and Subjects with Diabetes

Citation Author(s):
W. P. Samudika N.
Pathiratna
Department of Electronic and Telecommunication, University of Moratuwa, Sri Lanka
H. M. Heshani T.
Munasinghe
Department of Electronic and Telecommunication, University of Moratuwa, Sri Lanka
K. R. H. M. D. Manjitha
Kularatne
Department of Electronic and Telecommunication, University of Moratuwa, Sri Lanka
K. Rahul
Jeyanthan
Department of Electronic and Telecommunication, University of Moratuwa, Sri Lanka
Asma M.
Naim
Department of Electronic and Telecommunication, University of Moratuwa, Sri Lanka
Saroj
Jayasinghe
University of Colombo, Sri Lanka
Anjula
De Silva
Ear Science Institute Australia, Australia
Submitted by:
Asma Naim
Last updated:
Wed, 04/23/2025 - 18:56
DOI:
10.21227/6583-ay07
Data Format:
*.jpg
License:
Creative Commons Attribution
34 Views
Categories:
Artificial Intelligence
Machine Learning
Image Processing
Biomedical and Health Sciences
Medical Imaging
Keywords:
Diabetes dataset
0
0 ratings - Please login to submit your rating.
ACCESS DATASET CITE SHARE/EMBED

Abstract 

This dataset includes conjunctival and retinal images collected from both diabetic and healthy individuals to support research on diabetes-related vascular changes. For each subject, eight conjunctival images (four per eye: looking left, right, up, and down) are provided. Subjects with diabetes additionally have corresponding left and right retinal fundus images. Metadata for diabetic participants includes classification into subgroups: diabetes only, diabetes with retinopathy, or diabetes with related complications such as hypertension. A subset of 20 conjunctival images, from subjects with diabetes, has been manually annotated to highlight blood vessels larger than or equal to 40 μm, facilitating the analysis of vascular tortuosity. This dataset enables the development and validation of automated image analysis techniques for early, non-invasive screening of diabetes and diabetic retinopathy through ocular imaging.

Instructions: 

Instructions for Using the Dataset:

  1. Dataset Structure:

    • The dataset is organized by subject ID.

    • Each subject folder contains:

      • Conjunctival Images: 8 images (4 per eye – looking left, right, up, and down).

      • Retinal Images (only for subjects with diabetes): 2 fundus images (left and right eyes).

    • A separate folder includes 20 manually annotated conjunctival images from diabetic subjects, highlighting vessels ≥40 μm.

  2. File Formats:

    • All images are provided in high-resolution JPEG format.

    • Annotations are included as binary masks.

  3. Usage:

    • Conjunctival images may be used for vessel segmentation, tortuosity analysis, and CNN-based model training.

    • Retinal images can be used as reference or comparison for evaluating vascular changes.

    • The annotated subset is suitable for supervised learning or validation tasks.

  4. Tools Recommended:

    • Python (e.g., OpenCV, PyTorch), MATLAB, or similar platforms for analysis and model development.

Funding Agency: 
Senate Research Council of the University of Moratuwa
Grant Number: 
SRC/LT/2020/37

Dataset Files

LOGIN TO ACCESS DATASET FILES

Documentation

AttachmentSize
File README2.27 KB

 

QUESTIONS?