Diabetic Retinopathy, The Automated of Detection of Retinal Fundus Images with Probabilistic Neural Networks (PNN)

Abstract

Diabetic retinopathy and normal retinal diseases need a detection approach, in this case, to make the decision to diagnose diabetic retinopathy through retinal fundus images. The application aims to identify diabetic retinopathy through fundus imagery using Probabilistic Neural Network. The method we use involves the Probabilistic Neural Network (PNN) in the process of testing image data through the retina fundus. The results of this study have achieved detection of the accuracy of recognition of the range in reading images through the retina fundus of 88.9%, and The estimation of σ≥ 0.8 is the best benefit of smoothing boundaries to recognize diabetic retinopathy by utilizing the Probabilistic Neural Network, this shows the detection of research is quite high.