Team Maka
Diabetic retinopathy or diabetic eye disease is caused by diabetes mellitus which manifests itself in the eye retina. Diabetic eye disease is one of the most frequent causes of complete blindness in many developed countries. The detection of retinal pathologies became much easier using automated retinal image analysis whereas other methods like dilation of eye pupil is time consuming and patient has to suffer for some time. Diabetic retinopathy occurs when high blood glucose damages the small vessels that provides nutrients and oxygen to the retina. This paper focuses on automated computer-aided detection of diabetic retinopathy (DR) using features drawn from output of different retinal image processing algorithms, like diameter of optic disk, lesion specific (microaneurysms exudates), image level (pre-screening, AM/FM, quality assessment). These features are then used in an ensemble machine learning system comprising of different learning algorithms like alternating decision tree, adaBoost, Naïve Bayes, Random Forest and SVM.[1] The characteristic features extracted by anatomical part recognition algorithms and lesion detection are used to classify images. An ensemble of classifiers then uses these features to classify the whether an image has diabetic retinopathy or not. The most important features are the exudates which provide information about diabetic retinopathy in early stages. The major cause of exudates is the leakage of protein and lipids into the retina through damaged blood vessels. So, an ensemble based machine learning techniques are used to detect presence of diabetic retinopathy in an image. Source :IEE
Our approach was to built a model to detect the level of Diabetic Retinopathy with the help of machine learning and Deep learning.