Performance Analysis of Convolution Neural Network in Multiple Eye Disease Detection

Abstract

Age-related eye disease such as age-related macular degeneration, newlinecataracts, diabetic retinopathy and glaucoma are the leading causes of vision newlineloss in the elderly population. These diseases lead to death if not diagnosed at newlinean earlier stage. The only solution for this type of medical issues is early newlinedetection of these eye diseases. Despite the fact that machine learning (ML) newlinehas offered solutions for age-based severity categorization, electronic record newlinedatabase analysis, and automated early-stage prediction, several Machine newlinelearning algorithms are developed for the diagnosis and detection of diseases. newlineWhen it comes to databases, samples must always be well-structured and newlinebalanced, which is impossible for a complex database to provide. Additionally, newlinewhen compared to another current functioning model, the incidence based on newlineearlier predicted performance was significantly lower. All raw databases are newlinecurrently used for the machine learning subset known as quotdeep learning,quot which newlineis a well-known state of the art. Convolutional Neural Network (CNN)is one of newlinethe most used image processing methods in deep learning. The CNN deep newlinelearning method is totally proposed in research work aimed at picture newlinecategorization and precise prediction. comparing different data that have been newlinedownloaded from authentic medical image archives like ODIR. The main focus newlineof the research is the prediction of multiple eye diseases at dan earlier stages newlineutilising Deep Learning techniques and frameworks, which aids in the diagnosis newlineof patients who can be treated without progressing to the most severe stage of newlineage-related eye diseases. To begin with, a combined model of CNN-SVM for newlinethe classification of Diabetic retinopathy from the fundus images collected from newlinethe online database. Additionally, the second work is based on the hybrid model newlineof Adaptive Mutation Swarm Optimization which is used for the segmentation newlineof the diseased image attributes and Regression Neural Network as a classifier newlineto predict the multiple eye ds