International Journal of Engineering and Technology

Retinopathy of prematurity (ROP) is a disease that leads to blindness if the diagnosis is not made on time. Designing an automatic and intelligent system for early diagnosis can prevent the blindness in infants.However, the blood vessels in the retina of infants is less clear, hence the correct diagnosis of vessels is very difficult. In this study, first the pre-processing stage removes the noise of images, after that the images are divided into 16 × 16 windows and Gabor filter is applied to each sub-window. Energy and entropy are calculated for each sub-window, and finally these features are sent into a neural networkto determine whether the sub-window has a vessel or not.By counting the number of sub-windows containing vessel, the prematurenessand completeness of retina is determined. In this method, the mean accuracy of images that have been classified is 80.88% respectively.