Two Stages Approach for Detection and Classification of Retinal Images Using Deep Neural Network and Deep-wavelet

Abstract

Retina is the deepest layer in the inner surface regarding the eyeball and the
sensory layer of the eye. This sensitive part of the eye is susceptible to several
diseases. Realizing that the state regarding the retina represents one of the
primary causes of severe vision loss and blindness globally, retinal disease is
receiving major attention. Early diagnosis related to retinal pathology is now
based on an analysis of the geometric features of retinal blood vessels, including
branch lengths, widths, angles, tortuosity, branching patterns, and vessel
diameters. Various diseases can be diagnosed by specialists with using such
features. Since the year 1982, computer science has demonstrated its ability to
effectively contribute to the diagnosis and detection regarding disease for
biomedical sciences. The two stages of the method we present in this research
are for the classification and detection of diseases connected to the eyes. Level
one uses a parallel convolution neural network (CNN) for detecting the presence
of any disease. The Ocular Disease Intelligent Recognition Dataset and the
Ocular Disease Intelligent Recognition Dataset are two datasets used in level
two for classifying four eye-related diseases with the use of Deep Wavelet as a
technique for feature extraction (FE). We have decided that our approach
performs well and produces good results based on the comparative estimate