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 Video

Updated: Oct 21, 2025

Retinal Pathophysiological Evaluation in a Rat Model
09:11

Retinal Pathophysiological Evaluation in a Rat Model

Published on: May 6, 2022

4.8K

Explainable Diabetic Retinopathy Detection and Retinal Image Generation.

Yuhao Niu, Lin Gu, Yitian Zhao

    IEEE Journal of Biomedical and Health Informatics
    |September 8, 2021
    PubMed
    Summary

    This study introduces Patho-GAN, a novel deep learning method for interpretable medical diagnosis. It generates realistic retinal images to identify diabetic retinopathy symptoms, improving upon existing methods.

    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

    Plasmonic Modulation of the Upconversion Luminescence Based on Gold Nanorods for Designing a New Strategy of Sensing MicroRNAs.

    Analytical chemistry·2020
    Same author

    Vitamin D deficiency and metabolic syndrome in elderly Chinese individuals: evidence from CLHLS.

    Nutrition & metabolism·2020
    Same author

    Author Correction: Genome-wide identification of and functional insights into the late embryogenesis abundant (LEA) gene family in bread wheat (Triticum aestivum).

    Scientific reports·2020
    Same author

    Confirmation of the absence of local transmission and geographic assignment of imported falciparum malaria cases to China using microsatellite panel.

    Malaria journal·2020
    Same author

    circMET promotes NSCLC cell proliferation, metastasis, and immune evasion by regulating the miR-145-5p/CXCL3 axis.

    Aging·2020
    Same author

    Cyclic Peptide Extracts Derived From <i>Pseudostellaria heterophylla</i> Ameliorates COPD <i>via</i> Regulation of the TLR4/MyD88 Pathway Proteins.

    Frontiers in pharmacology·2020

    Area of Science:

    • Medical Imaging
    • Artificial Intelligence
    • Ophthalmology

    Background:

    • Deep learning models excel in disease classification but often lack interpretability.
    • Evidence-based medicine (EBM) principles, like Koch's Postulates, emphasize identifying causal factors.

    Purpose of the Study:

    • To enhance the interpretability of deep learning in medical diagnosis.
    • To develop a method for visualizing and understanding the features deep learning models use for predictions.
    • To generate synthetic medical images for data augmentation and symptom analysis.

    Main Methods:

    • Inspired by EBM, neuron activation patterns from a diabetic retinopathy (DR) detector were analyzed.
    • Novel pathological descriptors were defined using activated neurons to encode lesion information.

    More Related Videos

    Author Spotlight: Understanding Retinal Vessel Resilience and Disease Progression
    04:36

    Author Spotlight: Understanding Retinal Vessel Resilience and Disease Progression

    Published on: January 12, 2024

    1.3K
    Using Retinal Imaging to Study Dementia
    09:17

    Using Retinal Imaging to Study Dementia

    Published on: November 6, 2017

    21.8K

    Related Experiment Videos

    Last Updated: Oct 21, 2025

    Retinal Pathophysiological Evaluation in a Rat Model
    09:11

    Retinal Pathophysiological Evaluation in a Rat Model

    Published on: May 6, 2022

    4.8K
    Author Spotlight: Understanding Retinal Vessel Resilience and Disease Progression
    04:36

    Author Spotlight: Understanding Retinal Vessel Resilience and Disease Progression

    Published on: January 12, 2024

    1.3K
    Using Retinal Imaging to Study Dementia
    09:17

    Using Retinal Imaging to Study Dementia

    Published on: November 6, 2017

    21.8K
  • Patho-GAN, a new generative adversarial network, was proposed to synthesize retinal images based on these descriptors.
  • The ability to manipulate descriptors to control lesion characteristics was demonstrated.

    • Main Results:

    • Synthesized retinal images accurately reflected diabetic retinopathy symptoms.
    • Generated images were qualitatively and quantitatively superior to previous methods.
    • Patho-GAN demonstrated significantly faster image generation speeds compared to existing techniques.
    • The method allows for arbitrary control over lesion attributes in synthesized images.

    Conclusions:

    • The proposed approach enhances the interpretability of deep learning in medical diagnosis, specifically for diabetic retinopathy.
    • Patho-GAN offers a powerful tool for visualizing disease-specific features identified by AI.
    • The method's efficiency and control over image synthesis make it a promising solution for medical data augmentation.