Sistem Pengenalan Retina Menggunakan Self Organizing Map Untuk Mendeteksi Retinopati Diabetika

Marsiska Ariesta P, Iwan Setiawan W, Sri Mujiyono

Abstract

Eyes are important human sense. Diseases that damage many function of eye is diabetic retinopathy. Diabetic retinopathy is a microvascular complication that can occur in patients with diabetic and attacking vision function. Clinical symptoms of this disease is the emergence of mikroaneurisma which is swelling of blood vessels are microscopic and can be seen as reddish dots on the retina. The retina recognition process was done by taking the retina image data were processed using the Laplacian operator. Then do the feature extraction using Principal Component Analysis (PCA). PCA results of binary data is used as an input to the process of Neural Networks Self Organizing Map (SOM). The training process in order to make a decision about whether diabetic retinopathy or not . Results obtained with feature extraction Principal Component Analysis (PCA) with the variables, learning rate (a) = 0.6 , reduction of alpha (δ) = 0.5, threshold = 0.02 similarity and distance = 1x10^-15, has produced recognition rate by 85% for the best possible, and 50% for the worst possible.

Keyword : Retina Recognition, Principal Component Analysis, Self Organizing Map, Diabetic Retinopathy