A Computer-Aided Diagnoses System for Detecting Multiple Ocular Diseases Using Color Retinal Fundus Images

Abstract

Abstract Retinal images are acquired by a highly specialized camera called fundus camera. Retinal images play an essential role in diagnosing several ocular diseases, such as diabetic retinopathy (DR), atherosclerosis, and hypertension. These diseases can cause blindness if not classified accurately. The type of diseases can be specified by the density and appearance of the retinal blood vessels. Therefore, the blood vessels need to be segmented precisely before classifying the disease. We introduce a computer-aided diagnosis (CAD) system, which can detect multiple ocular diseases using multilabel classification by using the concept of problem transformation method. We combine the output of multiple class classifier by using multilevel support vector machine (MLSVM). Then, we made the comparisons between the output of our proposed system and the direct output of the multiclassifiers. The combination of some binary classifiers allows more extensions to improve the results. The CAD system of the multilabel classification system includes the acquisition of images, preprocessing, enhancement, segmentation of the regions that will be extracted, feature extraction and selection, classification, and evaluation.