Open AccessAutomatic Grading of Diabetic Retinopathy through Machine Learning
Author(s) -
Supriya Sangappa Kamatgi*,
Kalmeshwar N. Hosur
Publication year - 2020
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.i7148.079920
Subject(s) - naive bayes classifier , diabetic retinopathy , support vector machine , artificial intelligence , blindness , computer science , diabetes mellitus , decision tree , retinopathy , classifier (uml) , machine learning , grading (engineering) , medicine , bayes error rate , bayes classifier , pattern recognition (psychology) , optometry , engineering , civil engineering , endocrinology
Diabetes Retinopathy (DR) illness refers to a group of eye issues that can happen because of diabetes. It's the medical phenomenon within which a person’s retina is broken by the diabetes. During this stage the tinny blood vessels present within the retina get damaged due the high glucose level. This ends up damaging the tiny blood vessels within the retina leading to the loss of vision. Different complications occur due diabetes a number of them are upset, neuropathy, nephropathy, retinopathy, skin damages, hearing ailments. Globally, diabetic eye disease has become the fifth most common reason for blindness. Early identification of DR is important to forestall vision loss or blindness. In this paper the strategies like Naïve Bayes classifier, Bagged Decision Tree and Support vector machine are implemented and are used for the classification of data based on the training and testing datasets. The errors and accuracy of all the three classifiers are figured and the best among three is considered for the future application. This implementation is done in the MatLab software and results shows that the Bayes classifier gives the error 0.2, Bagged Decision Tree Classifier gives the error of 0.1 and the Support Vector machine gives the error of 0.04 is observed. Hence these observations shows that the Support Vector machines are good classifiers with the accuracy of 96%.