Identification of Diabetic Retinopathy (DR) using Image Processing
Abstract Diabetes appears in two varieties: Type-1 and Type-2. The former is chronic and can last for years together, whereas the latter can be cured if identified and treated at a premature stage. The symptoms of diabetes affecting the eyes appear very subtle and hence, identifying irregularities in retinal images is a demanding process for medical practitioners. Thus, there was a need to find a method to detect these abnormalities by observing the retinal images non-invasively. After going through research projects and recent developments in identifying DR, we found various techniques/strategies employed, their advantages and drawbacks followed by the objective of overall findings, and the importance of a good DR detection system. Our proposed method calls to attention the importance of early screening, using geometrical relations, multiple thresholding methods and usage of convolutional neural networks as means of overcoming the factors that stand as obstacles in timely detection.