Deep Learning - Based Diabetic Retinopathy Detection: A Survey on Deep Learning Architectures

Dr. Srinidhi G. A.

Abstract: A degenerative eye disorder Diabetic Retinopathy (DR), a condition that damages the retina occurs when blood glucose levels remain consistently high. It is necessary to detect the DR in the early stages to minimize future losses. Deep learning algorithms can analyze retinal fundus images, which helps in the early diagnosis of DR. This research looks at how deep learning approaches are used to recognize and classify retinal fundus images. In this paper, different deep - learning algorithms are studied to detect the presence of Diabetic Retinopathy. We examined different approaches in this study and evaluated the gaps in the literature about DR detection and classification, outlining their benefits and drawbacks. We also emphasize the challenges that lie ahead in developing robust, reliable, and effective deep - learning algorithms for diagnosing various Diabetic Retinopathy issues, as well as possible directions for future research.

Keywords: Deep learning, CNN, GoogleNet, AlexNet, VGGNet, RELU