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

Search research articles

Related Experiment Videos

Lesion Learning Network with Relation Aware Transformer for Diabetic Retinopathy Grading.

Hao Liang, Zhaoshui He, Zhijie Lin

IEEE Journal of Biomedical and Health Informatics

|April 10, 2026

Summary

A new deep learning model, Lesion Learning Network (LLNet), precisely grades Diabetic Retinopathy (DR) by focusing on lesion features. This AI approach improves diagnostic efficiency and accuracy for ophthalmologists, aiding early detection of DR.

Related Concept Videos

Diabetic Retinopathy

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

Single-cell RNA-seq reveals mRNAs and lncRNAs important for oocytes in vitro matured in pigs.

Reproduction in domestic animals = Zuchthygiene·2021

Same author

Effects of the histone acetylase inhibitor C646 on growth and differentiation of adipose-derived stem cells.

Cell cycle (Georgetown, Tex.)·2021

Same author

A Method for Estimating 24-Hour Urinary Sodium Excretion by Casual Urine Specimen in Chinese Hypertensive Patients.

American journal of hypertension·2021

Same author

<i>Scutellaria baicalensis</i> extract and baicalein inhibit replication of SARS-CoV-2 and its 3C-like protease <i>in vitro</i>.

Journal of enzyme inhibition and medicinal chemistry·2021

Same author

Frequent reassortment and potential recombination shape the genetic diversity of influenza D viruses.

The Journal of infection·2021

Same author

Evaluation of origanum oil, hydrolysable tannins and tea saponin in mitigating ruminant methane: In vitro and in vivo methods.

Journal of animal physiology and animal nutrition·2021

Area of Science:

Ophthalmology
Medical Imaging
Artificial Intelligence

Background:

Diabetic Retinopathy (DR) is a primary cause of irreversible blindness.
Early DR screening is difficult, impacting timely diagnosis and treatment.
Current automated DR grading methods struggle with intra-class variations and small lesion detection.

Purpose of the Study:

To develop a deep learning model for precise Diabetic Retinopathy grading.
To enhance the diagnostic efficiency and accuracy of ophthalmologists in DR detection.
To overcome challenges posed by intra-class variations and small lesions in DR grading.

Main Methods:

A Lesion Learning Network (LLNet) incorporating a Relation Aware Transformer was proposed.
A Lesion Information Extractor (LIE) identified and extracted fine-grained lesion features.

Related Experiment Videos

A Lesion Saliency Transformer (LST) enhanced perception of small lesions.

A Lesion Relation Aware Transformer (LRAT) modeled relationships between lesion conditions and severity grades.

Main Results:

LLNet achieved high accuracies on FGADR (82.7%), DDR (86.1%), and APTOS (87.2%) datasets.
The model demonstrated superior generalization, achieving 72.1% accuracy when trained on DDR and tested on EyePACS.
LLNet surpassed existing comparative methods in DR grading precision.

Conclusions:

The proposed LLNet effectively addresses challenges in DR grading, including small lesions and intra-class variations.
LLNet offers a robust and accurate automated solution for DR grading, improving diagnostic efficiency.
The adaptive lesion-learning strategy enhances data utilization and model generalization for DR screening.