Abstract:
This paper summarizes the different imaging techniques and methodologies used to perform classification of the different stages of a disease called diabetic retinopathy. In particular, it focuses on deep learning techniques to perform such detection in the fundus images of the patient’s eye. Diabetic retinopathy is a progressive disease that causes the patient to lose eyesight if not diagnosed and treated at an early stage. Ophthalmologists usually diagnose the patient of this disease by screening the retinal fundus images to look for lesions. But the inaccuracy of such diagnosis together with the delay between diagnosis and treatment motivated researchers to automate this process of diagnosis. Using neural networks to train the system on a set of training images, it is possible to make systems that are more accurate and faster than human experts.
