Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Nicotinamide Supplementation and Primary Open-Angle Glaucoma in Patients With Ocular Hypertension.

JAMA ophthalmology·2026

Same author

Risk and Predictors of Fellow-Eye Involvement Following Unilateral Retinal Artery Occlusion.

Ophthalmology. Retina·2026

Same author

Comparative Five-Year Risks of Systemic Complications with Biologic versus Conventional Therapy in Non-infectious Uveitis.

Ophthalmology·2026

Same author

Systemic Anemia and the Risk of Diabetic Macular Edema and Anti-VEGF Injections.

Ophthalmology. Retina·2026

Same author

Statin Intensity and the Risk of Noninfectious Uveitis.

Ophthalmology. Retina·2026

Same author

Public interest in retinal detachment in the United States: a Google Trends analysis.

International journal of retina and vitreous·2026

Search research articles

Related Experiment Video

Updated: Dec 14, 2025

Smartphone Fundus Photography

Smartphone Fundus Photography

Published on: July 6, 2017

Deep Learning Frameworks for Diabetic Retinopathy Detection with Smartphone-based Retinal Imaging Systems.

Recep E Hacisoftaoglu¹, Mahmut Karakaya¹, Ahmed B Sallam²

¹Dept. of Computer Science, University of Central Arkansas, Conway, AR, 72035, USA.

Pattern Recognition Letters

|July 25, 2020

Summary

Early detection of diabetic retinopathy (DR) is crucial for preventing vision loss. This study developed an accurate deep learning model using ResNet50 for smartphone-based retinal images, achieving 98.6% accuracy in detecting vision-threatening DR.

More Related Videos

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Using Retinal Imaging to Study Dementia

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

Related Experiment Videos

Last Updated: Dec 14, 2025

Smartphone Fundus Photography

Smartphone Fundus Photography

Published on: July 6, 2017

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Using Retinal Imaging to Study Dementia

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

Area of Science:

Ophthalmology
Medical Imaging
Artificial Intelligence

Background:

Diabetic Retinopathy (DR) poses a significant risk of vision loss and blindness if not detected early.
Annual eye exams are vital for diabetic patients to enable timely DR screening and prevent vision impairment.
Smartphone-based retinal imaging offers an affordable and accessible solution for DR screening in various settings.

Purpose of the Study:

To develop an automated DR detection model for smartphone-based retinal images using the ResNet50 deep learning architecture.
To evaluate the performance of different deep learning models (AlexNet, GoogLeNet, ResNet50) with transfer learning on diverse retinal image datasets.
To investigate the impact of field of view on DR detection accuracy in smartphone-based retinal imaging systems.

Main Methods:

Utilized transfer learning with AlexNet, GoogLeNet, and ResNet50 architectures.
Retrained models on multiple datasets (EyePACS, Messidor, IDRiD, Messidor-2) to assess data source effects.
Applied the optimized ResNet50 model to synthetic smartphone-based retinal images to evaluate real-world applicability.

Main Results:

The proposed ResNet50 model achieved high classification accuracy (98.6%) for vision-threatening DR detection.
Achieved excellent sensitivity (98.2%) and specificity (99.1%) with an AUC of 0.9978 on an independent test set.
Demonstrated improved DR detection accuracy using deep transfer learning with ResNet50 and publicly available datasets, considering field of view effects.

Conclusions:

Deep transfer learning with ResNet50 significantly enhances DR detection accuracy on smartphone-based retinal images.
The study highlights the feasibility of accurate DR screening using compact, affordable smartphone imaging systems.
The developed model provides a promising tool for early detection and management of diabetic retinopathy, even with limited training data and smaller fields of view.