Detection of Retinopathy of Prematurity using Convolution Neural Network

Abstract

Retinopathy of Prematurity (ROP) is an eye disease which leads to childhood blindness, it is most commonly found in premature babies, which is one of the largest preventable cause. The other name of ROP is retrolental fibroplasias (RLF), was first described in the mid-1900s. The use of Oxygen therapy in newborn babies is one of the major reason for cause of ROP. Due to this effect the role of supplemental oxygen in ROP has not been approved. Though there are many work has been undergone to detect ROP but the accurate method has not been identified. In ureti there will not be a normal development in blood vessels until the baby reach 40 weeks. The retina has not fully developed in premature babies. The retinal vessels may stop growing, or they may grow abnormally if retinal vascularization detached outside the uterus[1]. Due to abnormal development of blood vessels ROP occurs.ROP affects the new born babies vision. To detect this at earlier stage this work is proposed. In this project dataset of ROP and normal images are trained by using CNN algorithm and the test image is compared with available dataset and the result is provided as the test images is of normal or with ROP.