Applications of Deep Learning algorithms for retinal diseases diagnosis based on Optical Coherence Tomography imaging

Abstract

The accurate and early detection of retinal diseases is critical for effective treatment and prevention of vision loss. Optical coherence tomography (OCT) imaging has become an essential tool for non-invasive diagnosis of retinal diseases. In this paper, a deep learning approach for the automated diagnosis of retinal diseases using OCT images is presented. Specifically, five different deep learning models were evaluated, including DenseNet121, DenseNet169, DenseNet201, InceptionResNet, and a 12 Convolutional layers-based model, and compare their performance in terms of accuracy, AUC, f1-score, and loss. Our results demonstrate that the DenseNetl69 model achieved the highest accuracy (97%), AUC (0.997), f1-score (0.97) and a loss of 0.1 among the models evaluated. These results indicate the potential of deep learning models for the automated diagnosis of retinal diseases based on OCT images, and highlight the importance of continued research and development in this field to improve patient outcomes.

Authors

* External Author