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

Size estimation of mixed free-floating sub-resolution spherical scatterers utilizing adaptive scanning optical coherence tomography.

Optics continuum·2026

Same author

A cohort study of obese patients with moderate-to-severe psoriasis using biological medicines in Catalonia, Spain.

Dermatology reports·2026

Same author

Haploidentical transplant, gene therapy, and standard care in sickle cell disease: a cost-effectiveness analysis.

Blood·2026

Same author

Increasingly Reversible Na/Cl<sub>2</sub> and Li/Cl<sub>2</sub> Batteries.

Journal of the American Chemical Society·2026

Same author

Barriers to effective first-line treatment for women with iron deficiency.

Expert review of hematology·2026

Same author

Cost-effectiveness of iron deficiency screening in pregnancy across ferritin-based thresholds.

Blood advances·2026

Search research articles

Related Experiment Video

Updated: May 21, 2025

LED-Based In Vitro Screening for Assessing Photoactivable Molecules in Bacterial Photodynamic Inactivation

LED-Based In Vitro Screening for Assessing Photoactivable Molecules in Bacterial Photodynamic Inactivation

Published on: January 24, 2025

Improving Artificial Intelligence-based Microbial Keratitis Screening Tools Constrained by Limited Data Using

Daniel Wang¹, Bonnie Sklar², James Tian³

¹Duke University School of Medicine, Durham, North Carolina.

Ophthalmology Science

|March 21, 2025

Summary

Generative adversarial networks (GANs) can create synthetic slit-lamp images to improve artificial intelligence (AI) models for microbial keratitis (MK) screening, enhancing diagnostic accuracy with limited real data.

Keywords:

Artificial intelligence Cornea Generative adversarial network Microbial keratitis Slit-lamp photography

More Related Videos

Intense Pulsed Light for the Treatment of Dry Eye Owing to Meibomian Gland Dysfunction

Intense Pulsed Light for the Treatment of Dry Eye Owing to Meibomian Gland Dysfunction

Published on: April 1, 2019

Scanning Light Scattering Profiler SLPS Based Methodology to Quantitatively Evaluate Forward and Backward Light Scattering from Intraocular Lenses

Scanning Light Scattering Profiler SLPS Based Methodology to Quantitatively Evaluate Forward and Backward Light Scattering from Intraocular Lenses

Published on: June 6, 2017

Related Experiment Videos

Last Updated: May 21, 2025

LED-Based In Vitro Screening for Assessing Photoactivable Molecules in Bacterial Photodynamic Inactivation

LED-Based In Vitro Screening for Assessing Photoactivable Molecules in Bacterial Photodynamic Inactivation

Published on: January 24, 2025

Intense Pulsed Light for the Treatment of Dry Eye Owing to Meibomian Gland Dysfunction

Intense Pulsed Light for the Treatment of Dry Eye Owing to Meibomian Gland Dysfunction

Published on: April 1, 2019

Scanning Light Scattering Profiler SLPS Based Methodology to Quantitatively Evaluate Forward and Backward Light Scattering from Intraocular Lenses

Scanning Light Scattering Profiler SLPS Based Methodology to Quantitatively Evaluate Forward and Backward Light Scattering from Intraocular Lenses

Published on: June 6, 2017

Area of Science:

Ophthalmology
Medical Imaging
Artificial Intelligence

Background:

Microbial keratitis (MK) is a serious eye infection requiring prompt diagnosis.
Artificial intelligence (AI) models show promise for screening MK from slit-lamp photography (SLP).
Limited availability of diverse training data can hinder AI model performance.

Purpose of the Study:

To develop a novel generative adversarial network (GAN) model using limited data to supplement and improve an AI-based MK screening model.
To assess the quality and realism of GAN-generated synthetic SLP images.
To evaluate the impact of supplementing real training data with synthetic data on AI model performance for MK classification.

Main Methods:

A StyleGAN2-ADA GAN model was trained on real SLPs of healthy and MK eyes to generate synthetic images.
Synthetic image quality was assessed via a visual Turing test by cornea experts and quantitatively using Kernel Inception Distance (KID).
Two DenseNet121 AI models were trained: one with only real images and another with real images supplemented by GAN-generated synthetic images.

Main Results:

Cornea experts rated synthetic images as good quality, and both real and synthetic images depicted pertinent features for classification.
The MK screening model trained with supplemented synthetic data achieved a higher area under the receiver-operator characteristic curve (0.93) compared to the model trained with only real data (0.76).
Kernel Inception Distance (KID) analysis confirmed the realism and variation of the synthetic images.

Conclusions:

Supplementing limited real training data with synthetic data generated by GANs can significantly improve the performance of AI-based MK classification models.
GAN-generated synthetic SLPs are a viable tool for augmenting datasets in ophthalmology AI research.
This approach holds potential for enhancing the accuracy and robustness of diagnostic AI tools in clinical settings.