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

Real-World Practice in Secondary Stroke Prevention Following Embolic Stroke of Undetermined Source: Secondary Analysis of the CASPR Registry.

Neurology. Clinical practice·2026

Same author

Timing of Antithrombotic Therapy Among Patients With Hemorrhagic Transformation After an Ischemic Stroke.

Journal of the American Heart Association·2026

Same author

Probing Stage Transition Kinetics in Li-Graphite Intercalation Compounds by Time-Resolved In Situ Solid-State NMR via <sup>13</sup>C Labeling.

Journal of the American Chemical Society·2026

Same author

Three-year monitoring study of heavy metal fluxes and accumulation characteristics in mildly contaminated farmland soils of northern Guangdong, China.

Scientific reports·2026

Same author

Correction: Intravenous thrombolysis for acute central retinal artery occlusion: Protocol for a systematic review and individual participant data meta-analysis of randomized controlled trials.

PloS one·2026

Same author

Enzymatic preparation of luteolin-rich olive (Olea europaea L.) leaf extract and its inhibitory effects on colitis.

Bioorganic chemistry·2026

Search research articles

Related Experiment Video

Updated: Aug 3, 2025

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

Self-Supervised Equivariant Regularization Reconciles Multiple Instance Learning: Joint Referable Diabetic

Wenhui Zhu¹, Peijie Qiu², Natasha Lepore³

¹School of Computing and Augmented Intelligence, Arizona State University, AZ 85281, USA.

Proceedings of Spie--The International Society for Optical Engineering

|April 7, 2023

Summary

This study introduces a novel method combining self-supervised equivariant learning and attention-based multi-instance learning to accurately classify referable diabetic retinopathy (rDR) and segment lesions using only image-level labels.

Keywords:

Classification Diabetic Retinopathy Multiple Instances Learning Self-Supervised Method Weakly-Supervised Lesion Segmentation

More Related Videos

Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis

Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis

Published on: June 18, 2020

Using Retinal Imaging to Study Dementia

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

Related Experiment Videos

Last Updated: Aug 3, 2025

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

Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis

Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis

Published on: June 18, 2020

Using Retinal Imaging to Study Dementia

Using Retinal Imaging to Study Dementia

Published on: November 6, 2017

Area of Science:

Ophthalmology
Medical Imaging
Computer Vision

Background:

Distinguishing referable diabetic retinopathy (rDR) from non-referable DR is critical for patient care.
Existing large-scale datasets often lack pixel-level annotations, hindering detailed lesion analysis.
Developing algorithms for rDR classification and lesion segmentation from image-level labels is essential.

Purpose of the Study:

To develop and integrate algorithms for classifying rDR and segmenting lesions using only image-level labels.
To improve the accuracy of rDR classification by combining the strengths of self-supervised equivariant learning and attention-based multi-instance learning (MIL).

Main Methods:

Leveraged self-supervised equivariant learning and attention-based MIL for rDR classification and lesion segmentation.
Utilized MIL for differentiating positive (lesion) and negative (background) instances.
Employed a self-supervised equivariant attention mechanism (SEAM) for accurate, segmentation-level lesion localization.

Main Results:

Achieved an Area Under the Receiver Operating Characteristic Curve (AU ROC) of 0.958 on the Eyepacs dataset.
Demonstrated superior performance compared to current state-of-the-art algorithms.
Successfully integrated MIL and SEAM to enhance rDR classification accuracy.

Conclusions:

The proposed integrated approach effectively classifies rDR and segments lesions from image-level labels.
This method offers a significant advancement in automated diabetic retinopathy analysis.
The findings pave the way for more accurate and efficient screening tools in ophthalmology.