A diagnosis model for detection and classification of diabetic retinopathy using deep learning

• Published in: Network modeling and analysis in health informatics and bioinformatics (Wien) 2023-10, Vol.12 (1), p.37, Article 37
• Main Authors: Syed, Saba Raoof, M A, Saleem Durai
• Format: Article
• Language: English
• Subjects: Algorithms; Applications of Graph Theory and Complex Networks; Artificial neural networks; Bioinformatics; Blood vessels; Classification; Computational Biology/Bioinformatics; Computer Science; Datasets; Deep learning; Diabetes; Diabetes mellitus; Diabetes mellitus (non-insulin dependent); Diabetic retinopathy; Glucose; Health Informatics; Insulin; Machine learning; Methods; Microvasculature; Neural networks; Optimization techniques; Original Article; Retinopathy; Segmentation
• ISSN: 2192-6670; 2192-6662; 2192-6670
• DOI: 10.1007/s13721-023-00432-3

Diabetes mellitus (DM) is an immense progressive disease that affects the usage of blood glucose as energy, resulting in surplus glucose in the blood. If prolonged diabetes, it causes damage to both larger and smaller blood vessels, known as macrovascular and microvascular complications, respectively. The main objective of this paper is to develop an automated method for the detection, segmentation, and severity classification of type 2 diabetes mellitus (T2DM) microvascular complication Diabetic Retinopathy (DR) using the EyePACS dataset. An RU-Net (Residual U-Net) is proposed for segmentation, and a CCNN (Concatenated Convolutional Neural Network) for multi-class classification of DR. The proposed classification method recorded 0.9881% and 0.9683% accuracy for benchmark and real-time data. The result demonstrates that the proposed model is appropriate to assist physicians in the detection and classification of DR accurately and promptly.