Authors : Saranya P.; Muthulakshmi P.; Rishana M.; Vinothika S.; Sandhiya G.

Volume/Issue : Volume 11 - 2026, Issue 4 - April

Google Scholar : https://tinyurl.com/yhhs9ny9

Scribd : https://tinyurl.com/32avp43s

DOI : https://doi.org/10.38124/ijisrt/26apr914

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Diabetic Retinopathy is a serious eye disease caused by diabetes and is one of the leading causes of blindness worldwide. Early detection plays a crucial role in preventing vision loss; however, manual diagnosis requires expert ophthalmologists and significant time. In this paper, a machine learning-based system is proposed for the automated detection of diabetic retinopathy using retinal fundus images. The system involves image preprocessing, feature extraction, and classification using the XGBoost algorithm. The preprocessing stage enhances the quality of images and removes unwanted noise, while feature extraction captures important color, texture, and structural information from the retina. These features are used to train a classification model that can accurately distinguish between normal and affected cases. The system is evaluated using performance metrics such as accuracy, precision, recall, and F1-score. The results show that the proposed model achieves reliable performance with reduced computational complexity, making it suitable for deployment in real-world healthcare environments, especially in low-resource areas.

Keywords : Diabetic Retinopathy, Machine Learning, XGBoost, Image Processing, Feature Extraction, Medical Diagnosis.

References :

1. García, M., et al. (2023). Attention-Guided CNN for Early Diabetic Retinopathy. Using Messidor- 2 and DDR datasets, the model incorporates attention mechanisms for enhanced early lesion detection.
2. Hassan, M., et al. (2025). Adaptive Deep Learning for Diabetic Retinopathy Detection with Augmentation. Based on mixed public fundus datasets, applying adaptive deep learning with augmentation to enhance early-DR detection accuracy         .
3. Mohammed, A., et al. (2024). Multi-Scale Deep Network for Early Diabetic Retinopathy Detection. Built on Kaggle DR and APTOS datasets using multi-scale CNN with preprocessing for early-stage lesion sensitivity Ansarifar, J., Wang, L. and Archontoulis, S.V., 2021. An interaction regression model for crop yield prediction. Scientific reports, 11(1), p.17754.
4. Singh, R., et al. (2021). Deep Learning-Based Diabetic Retinopathy Detection. Using APTOS 2019 and Messidor datasets, the authors applied ensemble CNN models to improve accuracy and handle image quality variations.
5. Wang, H., et al. (2022). Transformer-Assisted Diabetic Retinopathy Grading. Utilized Kaggle DR and IDRiD datasets with a Vision Transformer + CNN fusion method for better feature representation and severity grading.
6. Zhang, Y., et al. (2024). Hybrid CNN and Vision Transformer Model for Diabetic Retinopathy. Trained on IDRiD and EyePACS datasets, this hybrid CNN + ViT approach improves classification using local and global features.
7. Kaggle “Diabetic Retinopathy Detection Dataset ”, Available: https://www.kaggle.com/datasets/tanlikesmath/diabeticretinopathy-resized