Detection of Retinopathy of Prematurity using ResNet Based Deep Features and Support Vector Machine Classifier

Retinopathy of prematurity (ROP) is a major problem that affects preterm babies all over the world. It can cause blindness in children, but it can be treated if it is found early and correctly. A way for diagnosing ROP disease called "deep learning" is suggested. First, the step of cleaning is done. Then, the suggested model uses the ResNet-50 feature map and gives it to a Support vector machine (SVM) to be categorized. So, metrics like accuracy, specificity, sensitivity, and precision are used to measure how well the model works. The results show that the suggested model is 98% more accurate than other methods. This shows that this method is better at predicting ROP in premature babies.