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: Jun 28, 2026

Functional Magnetic Resonance Imaging (fMRI) of the Visual Cortex with Wide-View Retinotopic Stimulation
07:11

Functional Magnetic Resonance Imaging (fMRI) of the Visual Cortex with Wide-View Retinotopic Stimulation

Published on: December 8, 2023

Clinician-Centered Evaluation Framework for Explainable AI Heatmaps in OCT-Based Retinal Disease Classification.

Eirini Maliagkani1, Ilias Georgalas1, Ioannis Datseris2

  • 11st Department of Ophthalmology, General Hospital of Athens "G. Gennimatas", National and Kapodistrian University of Athens, 11527 Athens, Greece.

Journal of Imaging
|May 26, 2026
PubMed
Summary

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

Neuroprotective Agents for Photoreceptor Rescue Following Experimental Retinal Detachment in Rodents: A Systematic Review.

Investigative ophthalmology & visual science·2026
Same author

RetinaDetachNet: Automated Deep Learning Quantification of Photoreceptor Cell Death for Neuroprotection Studies in Experimental Retinal Detachment.

Translational vision science & technology·2026
Same author

Focal Subthreshold Laser Photocoagulation of Eyelid Margin Vascularity in Telangiectatic Meibomian Gland Dysfunction.

Cornea·2026
Same author

MedScanGAN: Synthetic PET & CT Scan Generation Using Conditional Generative Adversarial Networks for Medical AI Data Augmentation.

Bioengineering (Basel, Switzerland)·2026
Same author

"Flapless" trocar-assisted intrascleral fixation technique for Carlevale IOL implantation without conjunctival opening: A novel approach.

European journal of ophthalmology·2026
Same author

Interpretable Machine Learning Identifies Key Inflammatory and Morphological Drivers of Intracranial Aneurysm Rupture Risk.

Bioengineering (Basel, Switzerland)·2026
This summary is machine-generated.

A new framework successfully selected clinically plausible explainable artificial intelligence (XAI) heatmaps for retinal optical coherence tomography (OCT) classification. The Token contRAST map (TRAST) method was rated highest by specialists for clinical relevance.

Area of Science:

  • Ophthalmology
  • Artificial Intelligence
  • Medical Imaging

Background:

  • Explainable Artificial Intelligence (XAI) is crucial for clinical adoption of AI in medical imaging.
  • Selecting clinically plausible XAI heatmaps for retinal Optical Coherence Tomography (OCT) classification remains a challenge.

Purpose of the Study:

  • To develop and validate a two-phase framework for selecting clinically plausible XAI heatmaps for OCT classification.
  • To compare the clinical plausibility of different XAI methods, including Token contRAST map (TRAST), Grad-CAM++, and Cosine-Grad Fusion Map (CGFM).

Main Methods:

  • A Swin Transformer model was trained and validated for OCT classification, achieving high diagnostic performance.
  • Phase 1 involved expert assessment of heatmap visual quality and anatomical plausibility.
Keywords:
clinical evaluationclinical plausibilityevaluation frameworkexplainable artificial intelligenceheatmapsmodel interpretabilityophthalmologyoptical coherence tomographyretinal disease classificationswin transformer

Related Experiment Videos

Last Updated: Jun 28, 2026

Functional Magnetic Resonance Imaging (fMRI) of the Visual Cortex with Wide-View Retinotopic Stimulation
07:11

Functional Magnetic Resonance Imaging (fMRI) of the Visual Cortex with Wide-View Retinotopic Stimulation

Published on: December 8, 2023

  • Phase 2 utilized a Likert scale for 21 specialists to evaluate the agreement between highlighted regions and model diagnosis.

    • Main Results:

    • The two-phase framework successfully narrowed down candidate XAI methods.
    • The Token contRAST map (TRAST) achieved the highest specialist ratings for clinical plausibility.
    • Gradient-weighted Class Activation Mapping (Grad-CAM++) performed better than Cosine-Grad Fusion Map (CGFM).

    Conclusions:

    • Effective XAI in OCT imaging requires structured expert evaluation beyond technical performance.
    • The proposed framework offers a practical approach for selecting clinically suitable XAI methods in ophthalmology.
    • TRAST demonstrates high potential for clinical application in OCT-based AI diagnostics.