Authors : R. Karthikeyan and P. Alli
Abstract: Now a days, in world wide the common diabetes related eye disease is a major issue for blindness. By utilization of image processing technique the abnormal detection of retinal images is obtained through regular screenings in the early stage itself. The accurate and early detection of issues is essential to protect the vision of patient in advance because it is the starting stage of the diseases. In order to avoid the issues a new method is proposed for identifying the abnormal and normal sign of retinal lesions. The abnormal detection is done by the automated medical system with hybrid classifier and optic disc coordinates. Finally, the image region features are formulated to binary map with its services based on certain properties like color, shape, etc. It detects the macula region of the images with ROI and the final output is obtained. The proposed method evaluation is done by the available digital retinal image databases and provides better performances with higher accuracy than the previous methods. The performances indicate the accuracy in terms of finding lesions and unusual signs in retinal images while maintaining correct classification of normal images.
R. Karthikeyan and P. Alli, 2016. Automated Detection of Retinal Lesions in Digital Retinal Images for Grading of Diabetic Retinopathy. Asian Journal of Information Technology, 15: 1015-1022.