Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

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

JAMA Ophthalmology-The Year in Review, 2025: Initiatives Hitting Their Stride.

JAMA ophthalmology·2026
Same author

Factors predicting fluid-free retina and complete polypoidal regression in polypoidal choroidal vasculopathy eyes receiving 2-mg aflibercept treatments.

Asia-Pacific journal of ophthalmology (Philadelphia, Pa.)·2026
Same author

JAMA Ophthalmology-Eye on AI.

JAMA ophthalmology·2026
Same author

Artificial Intelligence in Peer Review.

JAMA·2025
Same author

Diabetic retinal disease.

Nature reviews. Disease primers·2025
Same author

Myopia and high myopia trends in Chinese children and adolescents over 25 years: a nationwide study with projections to 2050.

The Lancet regional health. Western Pacific·2025
Same journal

The Role of Nrf2 in SIRT1-Mediated RGC Neuroprotection in Traumatic Optic Neuropathy.

Translational vision science & technology·2026
Same journal

Explicit Inclusion of Diabetes Mellitus Without Retinopathy Within Diabetic Retinopathy Prediction.

Translational vision science & technology·2026
Same journal

Preclinical Safety and Feasibility Study of Line-Field Confocal Optical Coherence Tomography for Ophthalmology Applications.

Translational vision science & technology·2026
Same journal

Pathogenicity Analysis of Two Novel CRB1 Mutations in Three Chinese Inherited Retinal Dystrophy Families and a Literature Review.

Translational vision science & technology·2026
Same journal

Gas-Lesion Contact and Postural Compliance After Vitrectomy With Tamponade: A Continuous Monitoring and 3D Quantitative Analysis.

Translational vision science & technology·2026
Same journal

Automated Deep Learning Quantification of Avascular Area and Intravitreal Neovascularization in Retinal Flatmounts of Rodent Oxygen-Induced Retinopathy Models.

Translational vision science & technology·2026
See all related articles

Related Experiment Video

Updated: Nov 5, 2025

Using Retinal Imaging to Study Dementia
09:17

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

21.9K

Addressing Artificial Intelligence Bias in Retinal Diagnostics.

Philippe Burlina1,2,3, Neil Joshi1, William Paul1

  • 1Applied Physics Laboratory, The Johns Hopkins University, Baltimore, MD, USA.

Translational Vision Science & Technology
|May 18, 2021
PubMed
Summary
This summary is machine-generated.

Generative AI methods can reduce bias in diabetic retinopathy diagnosis by creating synthetic images, improving diagnostic accuracy for all populations.

More Related Videos

Author Spotlight: Ex Vivo OCT-Based Multimodal Imaging of Human Donor Eyes for Research into Age-Related Macular Degeneration
10:14

Author Spotlight: Ex Vivo OCT-Based Multimodal Imaging of Human Donor Eyes for Research into Age-Related Macular Degeneration

Published on: May 26, 2023

3.8K
Techniques for Processing Eyes Implanted With a Retinal Prosthesis for Localized Histopathological Analysis
12:01

Techniques for Processing Eyes Implanted With a Retinal Prosthesis for Localized Histopathological Analysis

Published on: August 2, 2013

20.1K

Related Experiment Videos

Last Updated: Nov 5, 2025

Using Retinal Imaging to Study Dementia
09:17

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

21.9K
Author Spotlight: Ex Vivo OCT-Based Multimodal Imaging of Human Donor Eyes for Research into Age-Related Macular Degeneration
10:14

Author Spotlight: Ex Vivo OCT-Based Multimodal Imaging of Human Donor Eyes for Research into Age-Related Macular Degeneration

Published on: May 26, 2023

3.8K
Techniques for Processing Eyes Implanted With a Retinal Prosthesis for Localized Histopathological Analysis
12:01

Techniques for Processing Eyes Implanted With a Retinal Prosthesis for Localized Histopathological Analysis

Published on: August 2, 2013

20.1K

Area of Science:

  • Ophthalmology
  • Artificial Intelligence
  • Medical Imaging

Background:

  • Diabetic retinopathy (DR) diagnosis using deep learning systems (DLSs) can be affected by training data imbalances and domain generalization.
  • AI bias may arise from disparities in retinal image pigmentation, particularly affecting darker-skin individuals.
  • Existing DLSs may exhibit performance disparities across different demographic groups.

Purpose of the Study:

  • To evaluate generative methods for mitigating AI bias in diabetic retinopathy diagnosis.
  • To address performance disparities caused by training data imbalance and domain generalization in AI diagnostic tools.
  • To investigate the use of synthetic data generation for debiasing AI algorithms in ophthalmology.

Main Methods:

  • The study utilized the Kaggle EyePACS dataset, modifying it to create an artificial scenario of data imbalance and domain generalization.
  • A baseline diagnostic DLS was compared against DLSs trained with data augmented by generative models.
  • Generative models were employed to create synthetic fundus images, specifically addressing underrepresented groups (DR-referable darker-skin individuals).

Main Results:

  • The baseline DLS showed a significant accuracy disparity between lighter-skin (73.0%) and darker-skin (60.5%) individuals.
  • Generative methods successfully reduced the accuracy gap, achieving 72.0% for lighter-skin and 71.5% for darker-skin individuals.
  • The delta in accuracy between subpopulations decreased from 12.5% to 0.5% after applying generative debiasing techniques.

Conclusions:

  • Data imbalance and domain generalization can lead to significant AI diagnostic disparities.
  • Novel generative methods for synthetic fundus image creation show promise in debiasing AI diagnostic tools.
  • These AI advancements offer potential applications for improving fairness in DR diagnostics and other ophthalmic DLSs.