Deep learning based automated diagnosis of Ocular Toxoplasmosis in fundus images using convolutional neural network

Abstract

Ocular toxoplasmosis (OT) is often diagnosed by a specialist by the examination of fundus images of the eye. While deep learning is commonly used to process and identify diseases in medical images, ocular toxoplasmosis (OT) diagnosis has not received much attention up to this point.. We created and applied an effective Convolutional Neural Network (CNN) model that can accurately detect and classify Ocular Toxoplasmosis (OT) photos into four different groups: healthy, active, inactive, active-inactive. Later on, except healthy, three other classes turned to be an one class which is unhealthy. We created and applied an effective Convolutional Neural Network (CNN) model that can accurately detect and classify Ocular Toxoplasmosis (OT) photos into two different groups which are Healthy and Unhealthy. We claimed a proposed model that can accurately recognize and distinguish between the OT pictures on binary classes. In order to demonstrate the effectiveness of our customized Convolutional Neural Network (CNN) model, we employed four pre-trained models (VGG-16, VGG-19, MobileNet, ResNet50) and evaluated them using the same dataset. Our proposed custom model, along with four pretrained CNN architectures, demonstrates similar performance on the available dataset in terms of accuracy, precision, recall, and f1 score, as evaluated in this research. The proposed model shows a 95% accuracy rate. The CNN model recommended for diagnosing retinal disorders outperforms all previously utilized model.