VOL. 10, NO. 13, JULY 2015 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences © 2006-2015 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com 5631 AN AUTOMATIC GRADING SYSTEM OF SEVERITY LEVEL FOR DIABETIC RETINOPATHY USING CNN CLASSIFIER Mahendran Gandhi 1 and R. Dhanasekaran 2 1 Department of Electronic and Communication Engineering, Syed Ammal Engineering College, Ramanathapuram, India 2 Department of Electrical and Electronic Engineering, Syed Ammal Engineering College, Ramanathapuram, India E-Mail: vishnumahendran2010@gmail.com ABSTRACT The human eye is an organ that reacts to light and has several purposes. As a sense organ, the human eye allows vision. The retina is a light-sensitive layer at the back of the eye that covers about 65% of its interior surface. Rodand cone cells in the retina allow conscious light perception and vision including colour differentiation and the perception of depth. A disease called diabetic retinopathy which is affected in the retina to the diabetic patients suffering with the diabetes more than 10 years. Diabetic Retinopathy (DR) is the utmost common diabetic eye disease and a leading reason of blindness. It is caused due to injury in the blood vessels of the retina. The brittle blood vessels may enlarge and leak fluid which is called as exudates. Exudates are fluids, cells, or other cellular substances that are slowly discharged from blood vessels usually from inflamed tissues. It can be a clear fluid which is composed of serum, fibrin, and white blood cells. Diabetic retinopathy is identified by the ophthalmologists by the dilation method. The dilation method is meant by pouring eye drops into the patient’s eye and it causes irritation to the patients. In order to protect the patients from the irritation and blindness, an automatic method is developed to detect these exudates and to measure the severity of the diabetic retinopathy. The exudates are detected by using the JSEG algorithm and the severity level grading is done by using Cascade Neural Network (CNN) classifier algorithm. Keywords: diabetic retinopathy, exudates, principle component analysis, cascade neural network classifier. INTRODUCTION Diabetes is a major issue in our life caused due to increase in the sugar level content of our body because of lack of insulin. High blood sugar (glucose) enhances the risk of eye troubles from diabetes. In fact, diabetes is the leading reason of blindness in adults of ages 20 to 74. For diabetes, this high blood sugar causes the lens of the eye to enlarge, which changes the observing ability of the eye. To correct this category of eye troubles, the glucose level should be controlled. It may endureas long as three months after the glucose level is well controlled for the vision to completely get back to normal. Blurred vision may also be a sign of more dangerous eye problem with diabetes. The three major eye problems that can arise for the people with diabetes are cataracts, glaucoma, and retinopathy. Diabetic eye disease signifies to a group of eye problems which people with diabetes may face as a difficulty of diabetes. All can cause severe vision loss or even blindness. Diabetic retinopathy is the utmost ordinary diabetic eye disease steering to the vision loss. It is caused by variations in the blood vessels of the retina. For certain people with DR, blood vessels may enlarge and leak fluid. For some other people, abnormal new blood vessels nurture on the surface of the retina. The retina is the perceptive tissue at the rear of the eye. A healthy retina is essential for good vision. Blood vessels harmed from DR can cause vision loss in two modes: Fragile, abnormal blood vessels can arise and oozeblood into the centre of the eye so that the vision is blurred. This is called proliferative retinopathy and is the most progressed stage of the disease. Fluid can ooze into the center of the macula which is the part of the eye where sharp and straight-ahead vision occurs. The fluid causes the macula swelling to blur the vision. This condition is known as macular edema. It can happen at any stage of DR, even though it is more likely to occur as the disease proceeds. About half of the people with proliferative DR also have macular edema. Diabetic retinopathy has four stages: 1. Mild Non-proliferative Retinopathy- This is the earliest stage, where them icroaneurysms occur. They are small swelling that may break and allow blood to leak into nearby tissue. 2. Moderate Non-proliferative Retinopathy- As the disease progresses, some of the blood vessels that nurture the retina are clogged. 3. Severe Non-proliferative retinopathy-Many of the blood vessels are clogged, denying several areas of the retina with their blood supply. These areas of the retina deliver signals to the body to grow new blood vessels for sustenance. 4. Proliferative Retinopathy –This is the advanced stage, where the signals sent by the retina for sustenance cause the growth of new blood vessels. These additional blood vessels are abnormal and fragile. They grow near the retina and near the surface of the clear, vitrifiedgel filled inside of the eye. These blood vessels do not cause any symptomor vision loss. However, they have thin, breakable walls. If they oozeblood, dangerous vision loss and even blindness can result. The premature signs of DR are oozing blood vessels, retinal enlargement (macular edema), pale fatty sedimentson the retina–signs of oozingblood vessels, harmednerve tissue and any modifications to the blood vessels. DR can be noticed from dilated retinal images. Dilated images are taken by dropping chemical solution into patients’ eye. This permits the ophthalmologists to see