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

Contrastive explainable AI in healthcare: a systematic review of trends, benefits, research gaps, and future directions.

BMC medical informatics and decision making·2026
Same author

Progressive growing of patch size: Curriculum learning for accelerated and improved medical image segmentation.

Medical image analysis·2026
Same author

Integrating Participatory Social Innovation Into Requirements Engineering for AI Health Care Solutions: Case Study.

Journal of medical Internet research·2026
Same author

Big Data and Trustworthy AI for Heart Failure: A Review.

Circulation. Heart failure·2026
Same author

Bridging gaps in youth mental health care: YOUTHreach-a comprehensive European strategy.

European child & adolescent psychiatry·2026
Same author

Beyond benchmarks of IUGC: Rethinking requirements of deep learning method for intrapartum ultrasound biometry from fetal ultrasound videos.

Medical image analysis·2026

Related Experiment Video

Updated: Aug 26, 2025

Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
13:44

Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns

Published on: August 30, 2013

43.0K

Domain generalization in deep learning based mass detection in mammography: A large-scale multi-center study.

Lidia Garrucho1, Kaisar Kushibar1, Socayna Jouide1

  • 1Artificial Intelligence in Medicine Lab (BCN-AIM), Faculty of Mathematics and Computer Science, University of Barcelona, Gran Via de les Corts Catalanes 585, Barcelona, 08007, Barcelona, Spain.

Artificial Intelligence in Medicine
|October 7, 2022
PubMed
Summary

Deep learning for breast cancer detection struggles with domain generalization. This study introduces a novel training pipeline that improves performance across diverse clinical settings, outperforming existing methods.

Keywords:
Breast cancerData augmentationDigital mammographyDomain generalizationTransfer learningTransformer-based detection

More Related Videos

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
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

612

Related Experiment Videos

Last Updated: Aug 26, 2025

Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
13:44

Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns

Published on: August 30, 2013

43.0K
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
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

612

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Oncology

Background:

  • Deep learning (DL) shows promise for computer-aided detection (CAD) in breast cancer.
  • A key challenge is the poor domain generalization of artificial neural networks (ANNs) in varied clinical settings.
  • Domain shift in digital mammography hinders the reliable deployment of DL-based CAD systems.

Purpose of the Study:

  • To explore domain generalization of DL methods for mass detection in digital mammography.
  • To analyze sources of domain shift in a large-scale, multi-center setting.
  • To develop and evaluate a single-source training pipeline for improved domain generalization.

Main Methods:

  • Compared eight state-of-the-art DL detection methods, including Transformer models, trained on a single domain and tested on five unseen domains.
  • Developed a novel single-source mass detection training pipeline to enhance domain generalization without new domain data.
  • Conducted an in-depth analysis of covariate shifts impacting detection performance.

Main Results:

  • The proposed single-source training workflow demonstrated superior generalization compared to state-of-the-art transfer learning approaches in four out of five unseen domains.
  • The workflow effectively reduced domain shift attributed to variations in acquisition protocols and scanner manufacturers.
  • Identified key covariate shifts, including patient age, breast density, mass size, and malignancy, significantly affecting detection performance.

Conclusions:

  • The developed single-source training pipeline offers a promising solution for improving the domain generalization of DL-based breast cancer detection systems.
  • Understanding and mitigating covariate shifts is crucial for robust and reliable CAD systems in diverse clinical environments.
  • This study provides valuable insights and best practices for future research in DL domain generalization for medical imaging applications.