Retinal Image Based Examination of Glaucoma Using Deep Transfer Learning
F. Sangeetha Francelin Vinnarasi, Thaer Ahmad Abu-Saleem, Md. Tabil Ahammed, Manish Gupta, J. Bino, S. Prabha
Abstract
The eye is a very vital part of the body, and problems with it can lead to vision problems that are mild to severe. Image-guided approaches are widely used to do clinical-level eye health assessments. One of the most prevalent ways to look at the eyes is through retinal fundus imaging (RFI). This algorithm to sort the Normal/Glaucoma RFI database. There are four parts in this plan: gathering and tagging photos, using EfficientNet to find features, using a Butterfly Algorithm (BA) to improve features, and then using 3-fold cross validation to put them into two groups. The first step in the suggested strategy is to employ EfficientNet-model features to sort things and keep track of the outcomes. After that, the BDA is used to make the features better so that the most significant ones can be found. After that, a new features vector is made using serial features-fusion. The study's results suggest that this method can be 100% accurate when the BA optimised feature vector is utilised to do the classification task. In the future, this tool could be used to look at the RFI collected from hospitals.