Abstract
Diabetes mellitus is a form of diabetes with secondary microvascular complication leading to renal dysfunction and retinal loss also termed as diabetic retinopathy. Retinopathy is grave form of retinal disease. It is the leading cause of blindness in the world. Blockage of tiny minute retinal blood vessels due to the high blood sugar level is the reason why retinopathy leads to blindness or loss of vision. This study serves the purpose of deep learning-based diagnosis of Diabetic retinopathy using the fundus imaging of the eye. In this study architectures such as VGG 16 and VGG 19 are deployed in order to classify the images into 5 categories. The performance of the two models were compared. The highest accuracy is 77.67% when using the VGG 16 pre-trained model.
Highlights
Convolutional Neural Networks have been recently used in diagnosing of diabetic retinopathy through fundus imaging of the eyes
DR is classified into five stages which are non-diabetic retinopathy, mild form of DR, moderate form of DR, proliferate-DR and the last stage is the severe form of DR
Images were graded on a scale of 0 to 4 representing the stages of diabetic retinopathy
Summary
Convolutional Neural Networks have been recently used in diagnosing of diabetic retinopathy through fundus imaging of the eyes. Patients tend to turn up a stage where it is difficult to reverse the retinopathy, due to inadequate access to the detection and later eye care services. This calls for a noninvasive diagnostic system that is capable of detecting and garde them according to the stage described above. Algorithm to retrieve retinal vasculature to obtain and detect blood vessels was proposed by Tan et al Dekhil et al proposed a fine-tuned VGG-16 trained on Kaggle image database [3]. In this paper CNN architecture is proposed that classify images with high accuracy into the 5 controls (normal, mild, moderate, severe and proliferate). The performance was measured by obtaining the accuracy of the model
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Journal of Informatics Electrical and Electronics Engineering (JIEEE)
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
