Asian Journal of Information Technology

Year: 2014
Volume: 13
Issue: 11
Page No. 692 - 699

Adequate Pre-Processing Method to Estimate Neovascularization in Diabetic Retinopathy

Authors : K.G. Suma and V. Kavitha

Abstract: Fundus images are mainly used for screening the diabetic retinopathy, an eye disease which causes progressive damage to the retina due to the changes in hyperglycemia levels of the person. Neovascularization is the worst case of retinal abnormality of proliferative diabetic retinopathy where abnormal growth of new blood vessels grows in retinal part and the optic disc region of the eye due to the pressure built in the blood vessels. To analyze these blood vessels here the pre-processing method state initially, the neovascularization fundus image is split into R, G, B planes and hence each planes gets converted into gray scale image. Neovascularization blood vessels are twisted and thin nature, the Green plane of the image looks clearer than the Red and Blue plane of the fundus image. Another progression has been done as normalizing the input image by means of pre-specified mean and variance which increase the contrast of the image without changing the image features. After normalization the image will be subjected to split into R, G, B planes; here too the Green plane is selected which gives the confident level in finding the abnormal growth of blood vessels. Finally, the contrast scale difference between the two green planes of the images can be analyzed using pairwise Euclidean distance and brightness differences shown using bar chart. Performance outcome shows as the scale difference in 0.111 just increases the contrast of the image by suppressing the background gives the clear view of blood vessels used to analyze the neovascularization in fundus image to screen the DR for future research.

How to cite this article:

K.G. Suma and V. Kavitha, 2014. Adequate Pre-Processing Method to Estimate Neovascularization in Diabetic Retinopathy. Asian Journal of Information Technology, 13: 692-699.

Design and power by Medwell Web Development Team. © Medwell Publishing 2022 All Rights Reserved