Detection of Diabetic Retinopathy Using the Wavelet Transform and Feedforward Neural Network

摘要

The early detection of diabetic retinopathy plays a significant role in modern ophthalmology. This experimental work presents the detection of diabetic retinopathy images by the implementation of the wavelet transform and feedforward neural network. The wavelet transform segments blood vessels from the retinal images and the changes in retinal vasculature due to diabetic retinopathy are characterized by the vessel features. The vessel features of all the input images are evaluated by the wavelet-based retinal image analyzer to tabulate the input and target databases. The input and target matrices are then fed to the neural network to detect the diabetic retinopathy images from the set of normal and diabetic retina images.