A Complete Automated Diagnostic Tool for Non-Proliferative Diabetic Retinopathy

Authors

  • S. Sudha, A. Srinivasan, T. Gayathri Devi

Abstract

High blood sugar creates diabetes- related diseases. From nutriment, we get glucose, and insulin the hormone helps to take glucose into cells. If there are problems with insulin ,blood sugar levels will increase. It will cause eye disease, such as Diabetic Retinopathy (DR). In this automated system, fundus image is filtered using median filtering to remove noise, then it is enhanced by the contrast enhancement technique. Pre-processing is preferred to increase image fineness. It is followed by k-means image segmentation to detect non-proliferative stage  lesions such as Microaneurysms (Ma), Haemorrhages (He) & Exudates (Ex). Four major distinguishable features are extracted. It uses a Probabilistic Neural network classifier (PNN) for classification, and it detects the severeness of the eye disease. The result showed PNN classifier performed well for the detection, and classification of Microaneurysms (Ma), Haemorrhages and Exudates and got a high percentage of accuracy.

Downloads

Published

2020-05-10

Issue

Section

Articles