Automatic Prediction of Retinal Diseases using Image Processing Techniques

Abstract

Retina is a thin layer present behind the eye which is connected to the brain through optic nerve and it is
responsible for the visualization. Retinal disease is the most frequent cause of blindness for working age population.
Diabetes has become a new global challenge. If not diagnosed and treated in time, diabetes can encourage illness
related to the retina of human eyes that affects the retina and retinal structure in certain ways. The detection of such
abnormalities in the retina is called Diabetic Retinopathy (DR). Diabetic retinopathy is classified into two categories,
non-proliferative diabetic retinopathy (NPDR) and proliferative diabetic retinopathy (PDR). Age related macular
degeneration (AMD) is common cause of irreversible loss of central vision in elders. These two diseases are the most
common cause for the vision loss. Early detection helps the patients to aware of the seriousness of the disease and
prevents the blindness. This paper uses the process and knowledge of image processing to automatically predict DR
and AMD from fundus images of retina. The Pre-Processing stage equalizes the uneven lighting associated with
fundus images and removes noise present in the image. Segmentation stage partitions the image and converts it into
required form for feature extraction. Feature extraction stage extracts the various features from pre-processed and
segmented fundus image. In classification stage, the retinal diseases are classified as Normal, AMD, and DR (NPDR
or PDR).