Enhanced Diabetic Retinopathy Detection Using Self-Supervised Pretraining And Edge-Enhanced Feature Extraction Techniques
DOI:
https://doi.org/10.29284/8y5wym63Keywords:
Diabetic Retinopathy Detection, EfficientNetBO, Image Preprocessing, Deep Learning, Explainable AI, Grad-CAM Visualizations, Class Imbalance Handling, Feature Extraction, AUC Optimization, Low-Resource Deployment.Abstract
Diabetic Retinopathy (DR), a major complication of diabetes mellitus, remains a leading cause of blindness globally. Early detection and accurate grading are essential to prevent vision loss. This study provides a comprehensive comparative analysis of advanced deep learning methodologies for DR detection, focusing on EfficientNetB0 enhanced with preprocessing techniques such as CLAHE, Gaussian smoothing, circular cropping, and optic disc removal. Utilizing publicly available datasets (EyePACS, Messidor, and APTOS 2019), the proposed model demonstrated superior performance with an accuracy of 94.8%, sensitivity of 93.6%, specificity of 96.1%, and AUC of 97.8%. The experimental results highlight EfficientNetBO's advantage over other state-of-the-art models, including ResNet50 and a custom lightweight CNN. ResNet50 achieved an accuracy of 93.4% with an AUC of 96.5%, while the lightweight CNN, designed for resource-constrained environments, attained an accuracy of 92.1% and an AUC of 95.2%. Augmentation strategies further improved the generalization of EfficientNetBO, enhancing robustness to variability in image quality. Additionally, Grad-CAM visualizations ensured interpretability by highlighting critical retinal regions influencing predictions. This study also identified and addressed key challenges, such as variability in image quality, class imbalance, and limited interpretability, through innovative solutions like self-supervised learning, edgeenhanced feature extraction, and Bayesian hyperparameter optimization. Comparative results from existing works, including Gulshan et al. (2016) and Wang et al. (2021), underline the proposed approach's effectiveness in balancing sensitivity and specificity while achieving competitive AUC values. The findings validate EfficientNetB0 as a scalable and reliable model for DR detection, paving the way for its integration into clinical workflows. Future work will focus on enhancing interpretability, improving deployment in low-resource settings, and expanding the dataset diversity for greater generalization across populations.
References
1. Gulshan, Varun, et al. "Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs." JAMA, vol. 316, no. 22, 2016, pp. 2402-2410.
2. Zhou, Kun, et al. "A CNN-Based Approach with Image Enhancement for Diabetic Retinopathy Detection." International Journal of Artificial Intelligence, vol. 15, no. 2, 2017, pp. 50-65.
3. Liu, Cheng, et al. "Deep Learning Algorithms for Automated Detection and Classification of Diabetic Retinopathy." IEEE Transactions on Medical Imaging, vol. 38, no. 6, 2019, pp. 1605-1616.
4. Wang, Hong, et al. "Ensemble of Deep Learning Models for Robust Detection of Diabetic Retinopathy." Scientific Reports, vol. 11, no. 1, 2021, pp. 1-10.
5. Mohanty, Cheena, et al. "Using Deep Learning Architectures for Detection and Classification of Diabetic Retinopathy." Sensors, vol. 23, no. 12, 2023, pp. 5726.
6. Das, Dolly, Saroj Kumar Biswas, and Sivaji Bandyopadhyay. "Detection of Diabetic Retinopathy Using Convolutional Neural Networks for Feature Extraction and Classification." Multimedia Tools and Applications, vol. 82, no. 19, 2023, pp. 29943-30001.
7. Aziz, Tamoor, et al. "Deep Learning-Based Hemorrhage Detection for Diabetic Retinopathy Screening." Scientific Reports, vol. 13, no. 1479, 2023, pp. 1-10.
8. Alsubai, Shtwai, et al. "Quantum Computing Meets Deep Learning: A Promising Approach for Diabetic Retinopathy Classification." Mathematics, vol. 2023, no. 11, pp. 2008-2023.
9. Khan, Inam Ullah, et al. "A Computer-Aided Diagnostic System to Identify Diabetic Retinopathy, Utilizing a Modified Compact Convolutional Transformer and Low-Resolution Images to Reduce Computation Time." Biomedicines, vol. 11, no. 6, 2023, pp. 1566.
10. Sebastian, Anila, et al. "A Survey on Deep-Learning-Based Diabetic Retinopathy Classification." Diagnostics, vol. 13, no. 3, 2023, pp. 345.
11. Swamy, R. S., S. C. Kumar, and G. A. Latha. "An efficient skin cancer prognosis strategy using deep learning techniques." Indian Journal of Computer Science and Engineering (IJCSE) 12.1 (2021).
12. Kuna, Sri Lakshmi, and A. Prasad. "Deep Learning Empowered Diabetic Retinopathy Detection and Classification Using Retinal Fundus Images." International Journal on Recent and Innovation Trends in Computing and Communication, vol. 2022, pp. 117-119.
13. Sirisati RS, Kumar CS, Latha AG, Kumar BN, a Rao KS. An enhanced multi layer neural network to detect early cardiac arrests. In2021 5th International Conference on Electronics, Communication and Aerospace Technology (ICECA) 2021 Dec 2 (pp. 1514-1518). IEEE.
14. Dungrani, Dhruvin Rajesh, et al. "Detection and Classification of Diabetic Retinopathy Using Deep Learning." Proceedings of the IEEE International Conference on Nascent Technologies in Engineering, 2023.
15. Sirisati RS, Kumar CS, Venuthurumilli P, Ranjith J, Rao KS. Cancer sight: Illuminating the hidden-advancing breast cancer detection with machine learning-based image processing techniques. In2023 International Conference on Sustainable Communication Networks and Application (ICSCNA) 2023 Nov 15 (pp. 1618-1625). IEEE.
16. Gao, Zhentao, et al. "Diagnosis of Diabetic Retinopathy Using Deep Neural Networks." IEEE Access, vol. 7, 2018, pp. 3360-3367.
17. Bulla, Suneetha, et al. "A secure new HRF mechanism for mitigate EDoS attacks." International Journal of Ad Hoc and Ubiquitous Computing 40.1-3 (2022): 20-29.
18. Javier, Luis, et al. "Image Fundus Classification System for Diabetic Retinopathy Stage Detection Using Hybrid CNN-DELM." Big Data and Cognitive Computing, vol. 2022, no. 8, pp. 146-164.
19. SwamySirisati, R., Rao, M. S., & Thonukunuri, S. (2020, November). Analysis of Hybrid Fusion-Neural Filter Approach to detect Brain Tumor. In 2020 Sixth International Conference on Parallel, Distributed and Grid Computing (PDGC) (pp. 460-464). IEEE.
20. Srinivasa Rao, P. S. V., Mekala Srinivasa Rao, and Ranga Swamy Sirisati. "A Hybrid Clinical Data Predication Approach Using Modified PSO." Smart Computing Techniques and Applications: Proceedings of the Fourth International Conference on Smart Computing and Informatics, Volume 2. Singapore: Springer Singapore, 2021.
21. Swamy, Sirisati Ranga, et al. "Multi-Features Disease Analysis Based Smart Diagnosis for COVID-19." Comput. Syst. Sci. Eng. 45.1 (2023): 869-886.
22. Lee, Jung Hee, et al. "Assessment of MobileNet for Grading Diabetic Retinopathy." Journal of Clinical Imaging Science, vol. 11, no. 2, 2022, pp. 45-53.
23. Swamy, S. Ranga, et al. "Dimensionality reduction using machine learning and big data technologies." Int. J. Innov. Technol. Explor. Eng.(IJITEE) 9.2 (2019): 1740-1745.
24. Yau, Jonathan W. Y., et al. "Global Prevalence and Major Risk Factors of Diabetic Retinopathy." Diabetes Care, vol. 35, no. 3, 2012, pp. 556-564.
25. Kumar, C. Srinivasa, et al. "A smart recommendation system for medicine using intelligent NLP techniques." 2022 International Conference on Automation, Computing and Renewable Systems (ICACRS). IEEE, 2022.
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
