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 Experiment Videos

Prospective Data Curation Enables High-Performance Artificial Intelligence for Diabetic Retinopathy Screening in a

Cameron M Ashrafzadeh1, Milan Bahi1, Amira Mostafa2

  • 1Beetham Eye Institute, Joslin Diabetes Center, Boston, Massachusetts.

Ophthalmology Science
|June 1, 2026
PubMed
Summary

Related Concept Videos

Diabetic Retinopathy01:27

Diabetic Retinopathy

DefinitionDiabetic retinopathy is a microvascular complication of diabetes affecting the retinal blood vessels.Risk FactorsDiabetic retinopathy is present in almost all individuals with type 1 diabetes and more than 60% of those with type 2 diabetes after two decades of disease.The risk increases with poor glycemic control, hypertension, dyslipidemia, smoking, pregnancy, and puberty.Although cataracts and glaucoma are also more frequent in people with diabetes, retinopathy remains the leading...

You might also read

Related Articles

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

Sort by
Same author

Nanoparticle-based delivery of corn silk extract for the treatment of CCl₄-induced hepatotoxicity.

International journal of biological macromolecules·2026
Same author

Modifiable Factors Associated with Diabetic Retinal Disease and Retinal Neural Layer Thickness in Youth with Type 1 Diabetes.

Ophthalmology·2026
Same author

Improving the definition of glioblastoma resection borders with intraoperative cone-beam computed tomography.

Surgical neurology international·2026
Same author

Diabetic Retinopathy Severity on Ultra-Widefield Fluorescein Angiography versus Color Photography and Association with Risk of Disease Worsening.

Ophthalmology science·2026
Same author

Non-arteritic anterior ischaemic optic neuropathy incidence in placebo-controlled clinical trials of liraglutide or semaglutide.

The British journal of ophthalmology·2026
Same author

The Level of Knowledge and Attitude Regarding Family Planning Methods Among Married Men and Women in the United Arab Emirates.

Cureus·2026
This summary is machine-generated.

High-quality curated datasets can enable effective artificial intelligence as a medical device (AIaMD) models for diabetic retinopathy (DR) screening, even with minimal AI infrastructure. This approach yields reliable tools for equitable DR screening in resource-limited settings.

Area of Science:

  • Ophthalmology
  • Medical Imaging
  • Artificial Intelligence in Healthcare

Background:

  • Diabetic retinopathy (DR) screening is crucial for preventing vision loss.
  • Ultra-widefield (UWF) imaging offers comprehensive retinal views.
  • Developing effective AI for DR screening requires robust datasets.

Purpose of the Study:

  • To assess if high-quality, prospectively curated datasets can independently drive the development of effective AI as a medical device (AIaMD) models for DR screening.
  • To evaluate AI model performance for detecting more-than-mild DR (MTM) and diabetic macular edema (DME) using UWF fundus images.

Main Methods:

  • Collected 152,025 UWF color images from diabetic patients.
  • Trained two Convolutional Neural Networks (CNNs) using 26,232 images for MTM and DME detection.
Keywords:
Artifical IntelligenceDiabetic RetinopathyDiabetic macular edemaScreening

Related Experiment Videos

  • Validated models on an independent prospective dataset of 12,698 images.

    • Main Results:

    • The MTM model achieved high baseline performance (AUC 0.962) and maintained strong prospective results (AUC 0.949).
    • The DME model showed good baseline performance (AUC 0.879) and acceptable prospective results (AUC 0.821).
    • Gradient-weighted class activation mapping confirmed clinical relevance of AI model focus.

    Conclusions:

    • Rigorous data collection and quality control can yield high-performing AIaMDs with limited AI engineering resources.
    • Locally curated datasets can produce reliable, regulation-ready tools for equitable DR screening.
    • This approach is particularly valuable for resource-limited settings.