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

GUSL: A novel and efficient machine learning model for prostate segmentation on MRI.

Computers in biology and medicine·2026
Same author

Open-Geometry PET: Quantifying the Trade-off Between Time-of-Flight Resolution and Angular Coverage.

Physics in medicine and biology·2026
Same author

A scoping review of traditional and artificial intelligence methods in malaria diagnostics.

NPJ digital medicine·2026
Same author

A fully integrated sample-to-answer molecular diagnostic platform for rapid identification of four major <i>Aspergillus</i> species.

Frontiers in microbiology·2026
Same author

An Open Multi-Center Whole-Body FDG PET/CT Foundation Model for Tumor Segmentation.

ArXiv·2026
Same author

Influence of an AQP4 haplotype and sleep duration on early Alzheimer's disease.

Alzheimer's & dementia : the journal of the Alzheimer's Association·2026

Related Experiment Video

Updated: Sep 7, 2025

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
06:48

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images

Published on: January 7, 2019

9.0K

Unsupervised Black-Box Model Domain Adaptation for Brain Tumor Segmentation.

Xiaofeng Liu1, Chaehwa Yoo1,2, Fangxu Xing1

  • 1Gordon Center for Medical Imaging, Massachusetts General Hospital and Harvard Medical School, Boston, MA, United States.

Frontiers in Neuroscience
|June 20, 2022
PubMed
Summary

This study introduces a novel unsupervised domain adaptation (UDA) framework for medical image segmentation. It enables knowledge transfer without sharing sensitive source data or models, addressing privacy concerns in clinical settings.

Keywords:
MR imageblack-box modelbrain tumorknowledge distillationsegmentationunsupervised domain adaptation

More Related Videos

Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
04:25

Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies

Published on: December 15, 2023

2.8K
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: Sep 7, 2025

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
06:48

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images

Published on: January 7, 2019

9.0K
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
04:25

Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies

Published on: December 15, 2023

2.8K
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:

  • Medical Imaging
  • Machine Learning
  • Computer Vision

Background:

  • Unsupervised domain adaptation (UDA) facilitates knowledge transfer from labeled source domains to unlabeled target domains, crucial for overcoming data labeling challenges.
  • Existing UDA methods often require access to both source and target domain data, posing privacy risks and collaboration hurdles in clinical settings due to sensitive patient information.

Purpose of the Study:

  • To propose a practical UDA framework for medical image segmentation that operates with a black-box source model, eliminating the need for source data or white-box access.
  • To enhance UDA performance by incorporating a knowledge distillation scheme and unsupervised entropy minimization for improved target-specific representation learning and label confidence regularization.

Main Methods:

  • Developed a novel UDA framework utilizing a black-box segmentation model trained solely on source domain data.
  • Implemented a knowledge distillation strategy to progressively learn target-specific feature representations.
  • Applied unsupervised entropy minimization to regularize the confidence of predictions in the target domain, boosting adaptation performance.

Main Results:

  • The proposed framework successfully adapted models to new domains without requiring access to source data or model parameters.
  • Unsupervised entropy minimization demonstrated performance gains compared to UDA methods lacking this regularization.
  • Extensive validation across multiple datasets and deep learning backbones confirmed the framework's efficacy.

Conclusions:

  • The developed UDA framework offers a privacy-preserving and practical solution for cross-center collaborations in medical image analysis.
  • The integration of knowledge distillation and entropy minimization provides a robust approach for unsupervised domain adaptation in challenging clinical environments.
  • This work paves the way for more reliable and widely applicable AI tools in healthcare by addressing data sharing and privacy limitations.