Deep Transfer Learning Approaches to Predict Glaucoma, Cataract, Choroidal Neovascularization, Diabetic Macular Edema, DRUSEN and Healthy Eyes: An Experimental Review

Artificial intelligence (AI) has lately witnessed an age of tremendous expansion across several industries, including healthcare. In recent years, substantial advancements in AI, notably in machine learning (ML) and deep learning (DL), have had a considerable influence on traditional diagnostic and...

Full description

Saved in:
Bibliographic Details
Published in:Archives of computational methods in engineering 2023, Vol.30 (1), p.521-541
Main Authors: Kumar, Yogesh, Gupta, Surbhi
Format: Article
Language:English
Subjects:
Accuracy
Artificial intelligence
Cataracts
Decision making
Deep learning
Diabetes
Edema
Engineering
Eye diseases
Glaucoma
Health care
Health services
Imaging techniques
Machine learning
Mathematical and Computational Engineering
Medical imaging
Natural language processing
Object recognition
Review Article
Voice recognition
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Artificial intelligence (AI) has lately witnessed an age of tremendous expansion across several industries, including healthcare. In recent years, substantial advancements in AI, notably in machine learning (ML) and deep learning (DL), have had a considerable influence on traditional diagnostic and treatment procedures. Although DL has been widely used in image identification, voice recognition, and natural language processing, it is now beginning to influence healthcare. The principal imaging techniques for early detection and treatment of eye problems are now fundus digital photography and optical coherence tomography (OCT). The study predicts the presence of multiple eye diseases in the human eye. Multiple images of eye diseases like diabetic macular edema (DME) and choroidal neovascularization (CNV), DRUSEN, GLAUCOMA, NORMAL, and CATARACTS have been used to train and validate AI-based intelligence learning approaches. This study proposed multiple transfer learning models for the prediction of eye diseases. The deep transfer learning approaches used in the study are Basic CNN, Deep CNN, AlexNet 2, Xception, Inception V3, ResNet 50, and DenseNet121. The simulation results verify that the ResNet50 attained 98.9% validation accuracy and outperformed all the other approaches. Also, the Xception model performed well and achieved 98.4% accuracy. The training and validation loss achieved by the Xception model is 0.15 and 0.05, respectively. The best root mean squared error achieved by the Xception model is 0.22. Thus, the research aims to enhance clinical decision-making that has a variety of applications for ophthalmologists.
ISSN:1134-3060
1886-1784
DOI:10.1007/s11831-022-09807-7