Development of Computer-aided Conjunctival Hyperemia Screening System based on Eye Image using Machine Learning Algorithms

Abstract

Conjunctival Hyperemia is an eye disease, which in critical stage may leadto vision loss. The damage due to Hyperemia can be avoided, if detected and treated during early stages. There istremendousscope to develop automated Hyperemia diagnosis tool based on image dataset as training data to machine learning algorithms. In this proposed work, an attempt is made to develop such a system where feature vector is created using DCT and DWT. This dataset is used for training supervised learning algorithms namely like SVM, random forest and neural network. The work shows promising results with average 80% accuracy.