Diabetic retinopathy parameters detection using convolutional neural network

Diabetic Retinopathy is a disorder of the retinal blood vessels that occurs as a complication of diabetes mellitus.
It can be caused the risk of decreased vision function to permanent blindness if the patient can not recognize it earlier. Basic
Health Research conducted by the Ministry of Health stated that in almost all provinces in Indonesia, diabetic retinopathy
had increased. Therefore, an analysis is needed to recognize diabetic retinopathy according to its severity. In this study,
Convolutional Neural Network engineering was used to measure the accuracy of diabetic retinopathy detection. The
parameters used are stride filter and zero padding. The stride parameter is three, and zero padding is the same. Three images
were taken on the testing data, and the results obtained were 33.16% accuracy with Mild severity, 30.11% accuracy with
Moderate severity, and 39.79% accuracy with Normal severity