An Artificial Intelligence Driven Framework for Classification of Ophthalmic Images Using Convolutional Neural Network

• DOI: 10.2139/ssrn.4531893
• Publication Date: 2023-08-09T19:07:04Z
• Authors: Shagundeep Singh; Raphael Banoub; Harshal A. Sanghvi; Ankur Agarwal; K.V. Chalam; Shailesh Gupta; Abhijit S. Pandya

Early disease detection is emphasized within ophthalmology now more than ever, and as a result, clinicians innovators turn to deep learning expedite accurate diagnosis mitigate treatment delay. Efforts concentrate on the creation of systems that analyze clinical image data detect disease-specific features with maximum sensitivity. Moreover, these hold promise early patients common progressive diseases. DenseNet, ResNet, VGG-16 are among few Convolutional Neural Network (CNN) algorithms have been introduced being investigated for potential application ophthalmology. In this study, authors sought create evaluate novel ensembled CNN model analyzes dataset shuffled retinal color fundus images (RCFIs) from eyes various ocular (cataract, glaucoma, diabetic retinopathy). Our aim was determine (1) relative performance our finalized in classifying RCFIs according disease, (2) diagnostic serve screening test specific diseases retinopathy) upon presentation diverse manifestations. We found adding convolutional layers an existing resulted significantly increased 98% accuracy (p<0.05) including good binary cataract, retinopathy.