A Hybrid Approach for Diagnosing Diabetic Retinopathy in Retinal Fundus Images using an Adaptive Canny Edge Detection approach

• Published in: 2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT) pp. 1 - 6
• Main Authors: Manjula Devi, R, Kowsalya, M, Dinesh, P M, Ponmurugan, P, Berlin, A, Kavin Kumar, K, Sangeetha, M, Chandrasekaran, Gokul
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24-06-2024
• Subjects: Accuracy; Adaptation models; Blindness; Canny Edge Detection Introduction; Classification algorithms; Computational modeling; Diabetes; Diabetic retinopathy; Diseases; Image edge detection; Neural Networks; Retina; Training
• ISSN: 2473-7674
• DOI: 10.1109/ICCCNT61001.2024.10723952

Individuals diagnosed with diabetes are susceptible to developing diabetic retinopathy (DR), a degenerative eye condition that affects the retina. When blood vessels in the retina get damaged, it can cause permanent vision impairment or even blindness if the underlying condition is not addressed quickly enough. Early diagnosis and treatment of this disease can save a patient's eyesight and prevent blindness. Successful medications for DR are available, but people with diabetes need to be identified and monitored early on. In order to diagnose diabetic retinopathy, many physical examinations have to be conducted; however, these types of exams require more time to complete. Adaptive Canny Edge Detection Training using Adaptive Skipping technique (ACEDIAS) is used in this research to speed up the detection process of diabetic retinopathy by detecting the edge and boost diagnostic accuracy. The study team chooses to use a 2063image dataset compressed from Kaggle's library. The suggested ACEDIAS model is evaluated for its performance using the measures such as accuracy, recall, precision, and F1score. It has been demonstrated through simulation that the suggested ACEDIAS model outperforms the existing models.