Textural measurements and pattern recognition based diabetic retinopathy classification system of retinal images

Abstract

Diabetes is a serious disease that affects many vital organs of a human body, including the eyes. In diabetic patients, the variation of the sugar levels in the blood leads to several complications in the eyes. Around 80 percentage of the patients, inflicted by diabetes for more than 10 years have severe eye related pathological conditions such as retinopathy and maculopathy. Appropriate detection, diagnosis and treatment of eye related pathologies prevent the damage to the eye during the earliest stages of the diabetic disease. newlineThe advanced stage results patient s losing the vision. The retinal damage due to diabetes is termed as Diabetic Retinopathy (DR). Chronic diabetic retinopathy is a major threat to diabetic patients with its progression damaging the blood vessels of the retina and causing abnormalities in the macular region. The diabetic retinopathy if left untreated, can lead to blurred eye vision and blindness. The treatment of diabetic retinopathy involves detecting the presence of the disease in the form of micro aneurysms, haemorrhages and exudates in the retinal area. newlineThe first work, proposes an efficient method of glaucoma recognition at an early stage so as to minimize the damage of optic nerve head. The proposed system is based on the analysis of the instrumental parameters and other important risk factors for glaucoma development. The results obtained show that the used methods possess an enormous power that meets the specifications. Also the obtained results are compared with the performance of medical experts in detecting glaucomatous eyes from the same images used in the study. For further validation of the proposed system s performance, sensitivity, specificity and accuracy are calculated. newline