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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

3D OCT-Based Retinal Biomarker Analysis for Automatic Regional-Wise Characterization of Neurodegenerative Diseases.

Journal of imaging informatics in medicine·2026
Same author

Engineering and application of a recombinant Aspergillus niger CECT 2907 tannase to enhance the antioxidant activity of black tea.

Journal of the science of food and agriculture·2026
Same author

Unsupervised Deep Learning-based Keypoint Localization Estimating Descriptor Matching Performance.

IEEE journal of biomedical and health informatics·2026
Same author

DNA methylation biomarkers for cervical cancer risk prediction in HIV-positive Nigerian women.

International journal of cancer·2025
Same author

Pleomorphic Adenoma in the Setting of Triple-Negative Breast Cancer: A Case Report.

Clinical case reports·2025
Same author

Efficient clinical decision-making process via AI-based multimodal data fusion: A COVID-19 case study.

Heliyon·2024

Related Experiment Video

Updated: Aug 16, 2025

Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
07:12

Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss

Published on: April 11, 2025

507

Context encoder transfer learning approaches for retinal image analysis.

Daniel I Morís1, Álvaro S Hervella1, José Rouco1

  • 1Centro de Investigación CITIC, Universidade da Coruña, Campus de Elviña, s/n, 15071 A Coruña, Spain; Grupo VARPA, Instituto de Investigación Biomédica de A Coruña (INIBIC), Universidade da Coruña, Xubias de Arriba, 84, 15006 A Coruña, Spain.

Computers in Biology and Medicine
|December 26, 2022
PubMed
Summary

Deep learning for retinal image analysis faces data scarcity. This study introduces Context Encoder methods to improve transfer learning, effectively analyzing retinal structures with less labeled data.

Keywords:
Biomedical imagingContext EncoderDeep learningEye fundusSelf-supervised learningTransfer learning

More Related Videos

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

601
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.9K

Related Experiment Videos

Last Updated: Aug 16, 2025

Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
07:12

Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss

Published on: April 11, 2025

507
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

601
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.9K

Area of Science:

  • Biomedical image analysis
  • Deep learning in healthcare
  • Ophthalmology imaging

Background:

  • Deep learning models require substantial labeled data for effective training, which is challenging in biomedical imaging due to the need for expert annotation.
  • Data scarcity is a significant bottleneck in developing accurate deep learning models for retinal image analysis.
  • Transfer learning is a common strategy to address data scarcity, but its success relies on effective pre-training techniques.

Purpose of the Study:

  • To explore the Context Encoder paradigm for transfer learning in retinal image analysis.
  • To propose novel approaches for Context Encoder pre-training that accommodate full-resolution images and enhance retinal structure recognition.
  • To evaluate the effectiveness of these approaches in mitigating data scarcity for critical retinal imaging tasks.

Main Methods:

  • Application of the Context Encoder paradigm for pre-training deep neural networks.
  • Development of several approaches to process full-resolution retinal images within the Context Encoder framework.
  • Fine-tuning of Context Encoder pre-trained models for retinal vessel segmentation and fovea localization tasks.
  • Experimental validation using diverse public retinal image datasets.

Main Results:

  • The proposed Context Encoder approaches effectively mitigate the impact of data scarcity in retinal image analysis.
  • These methods demonstrate superior performance compared to previous alternatives for retinal structure recognition tasks.
  • Successful application in both retinal vessel segmentation and fovea localization, highlighting versatility.

Conclusions:

  • Context Encoder-based transfer learning offers a robust solution to the data scarcity problem in retinal image analysis.
  • The developed methods improve the recognition of crucial retinal structures, advancing automated diagnostic capabilities.
  • This work provides a foundation for more efficient and accurate deep learning applications in ophthalmology.