WSEAS Transactions on Biology and Biomedicine
Print ISSN: 1109-9518, E-ISSN: 2224-2902
Volume 11, 2014
Diagnosis of Diabetic Retinopathy by Extracting Blood Vessels and Exudates Using Retinal Color Fundus Images
Authors: ,
Abstract: Diabetic retinopathy (DR) is the damage caused by complications of diabetes to the retina. It is one of the leading causes of blindness across the world. Hence, an accurate, premature diagnosis of DR is an essential task because of its potentiality for reducing the number of cases of blindness across the globe. The main objective of our study was to develop a cost-effective computer-aided diagnostic system (CAD) in order to evaluate the performance of the system which automatically classifies images with pathologic features commonly found in DR. This study was performed on 60 South Indian subjects whose age ranged from 50-85 years. For all the subjects, digital images of size 640 x 480 were taken with a CARL ZEISS FF 450 plus Visupac fundus camera. The ground truth results were provided for the presence of pathological conditions such as micro aneurysms, exudates, hemorrhages. An SVM kernel classifier based CAD system was used to report the presence or absence of DR. The next step was the evaluation of the diagnostic capability of the proposed method in order to identify the subjects with DR by means of sensitivity, specificity and accuracy with respect to ground truth results. The proposed system has attained uppermost classification accuracy, reported so far by means of 5-fold cross validation analysis with the average sensitivity, specificity and accuracy values of 91.6%, 90.5% and 91.2% respectively. In conclusion, our findings suggest that the proposed CAD system would be a useful technique for cataloging the subject with DR.
Search Articles
Keywords: Fundus image, blood vessel, image processing, diabetic retinopathy, mathematical morphology, Support Vector Machine
Pages: 20-28
WSEAS Transactions on Biology and Biomedicine, ISSN / E-ISSN: 1109-9518 / 2224-2902, Volume 11, 2014, Art. #4