Comparative Analysis of Early Diabetic Retinopathy Detection: Enhanced Minimal CNN vs VGG Architecture on CLAHE- Preprocessed Retinal Images

Abstract

Ocular impairment is one of the prominent problems affecting middle-aged individuals due to uncontrolled blood sugar levels, commonly known as Diabetic Retinopathy (DR). The small abnormalities in the retinal capillaries, called microaneurysms and intra retinal bleeding, are the initial symptoms of Diabetic Retinopathy. Clinically recognizing diabetic retinal disease is a time-consuming and difficult process due to limitations in resources and experienced doctors. Early detection is crucial in avoiding the progression of Diabetic Retinopathy, highlighting the importance of an automated DR detection method to identify symptoms in its early stages. In this paper, researchers developed an unfamiliar framework known as Enhanced Minimal Convolutional Neural Network (EMCNN) to classify Mild-DR and No-DR ophthalmic photos using a binary classification process. The proposed new model EMCNN is compared with the migration learning method using the existing framework VGG16 and VGG19 in terms of precision and effectiveness. Before being sent across the network, the fundus pictures underwent preprocessing using the Contrast Limited Adaptive Histogram Equalization (CLAHE) tactic. EMCNN is an experimental model that enjoys a minimum number of layers and batch normalization to minimize the training effort. The EMCNN model achieved 94.89% accuracy using 3100 image dataset which is a remarkable improvement when compared with VGG architectures since the VGG architecture is trained with millions of images.