Early Detection and Classification of Diabetic Retinopathy using K-Means Colour Compression and Fuzzy Logic

Conference: International Conference on Soft Computing Applications in Wireless Communication
Author(s): Sumandeep Kaur, Daljit Singh Year: 2017
Grenze ID: 02.SCAWC.2017.1.517 Page: 243-251


Diabetic retinopathy is anailmentinitiateddue to diabetes mellitus. The longer an individual is suffering from\ndiabetes the more is the possibility of diabetic retinopathy. There are certain abnormalities that are caused by DR, like Microaneurisms,\nCotton wool spots, soft exudates and hard exudates. The cotton and soft exudates are the most primitiveindication\nof DR while Hard exudates and Hemorrhages are experienced in the advancedphases of the diabetic retinopathy. The aim of the\nprojected work is to effectively diagnose and categorize diabetic retinopathy. K-means compression technique is helpful in\ndecreasingcolor dimensions by clustering the fundus image in various areas of concern. These areas are then segmented out\nand analyzedover various region characteristics. Lastly, theclassification of fundus imagesis done with the help of\nknowledge-based (KB) fuzzy inference system (FIS) by using the effective characteristics of the region. The accuracyand\nsensitivity of the system arecalculatedto be 94.7% and 96% respectively.


SCAWC - 2017