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

Lung cancer organoids for functional precision oncology: from disease modeling to clinical decision support.

Frontiers in oncology·2026
Same author

Primary Cilia as Integrative Hubs of Metabolic Signaling in Type 2 Diabetes: Inter-Organ Evidence From Central, Peripheral, and Pancreatic Islet Tissues.

Journal of cellular physiology·2026
Same author

A pan-cancer single-cell atlas uncovers the role of sex hormones and chromosomes in sex-divergent reprogramming of the tumor microenvironment.

Cell communication and signaling : CCS·2026
Same author

Mechanotransduction in glioma stem cell fate determination: from niche mechanics to therapeutic vulnerability and state plasticity.

Frontiers in cell and developmental biology·2026
Same author

WAT-to-BAT communication facilitates the sustained activation of BAT thermogenesis during cold exposure.

Cell discovery·2026
Same author

Sti1 participates in the dynamics of protein aggregation triggered by glucose signaling in <i>Saccharomyces cerevisiae</i>.

Acta biochimica et biophysica Sinica·2026

Related Experiment Video

Updated: Oct 7, 2025

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

3.0K

Margin Preserving Self-Paced Contrastive Learning Towards Domain Adaptation for Medical Image Segmentation.

Zhizhe Liu, Zhenfeng Zhu, Shuai Zheng

    IEEE Journal of Biomedical and Health Informatics
    |January 6, 2022
    PubMed
    Summary

    This study introduces a new method for unsupervised domain adaptation in medical imaging, improving segmentation accuracy by focusing on class-specific feature alignment. The margin preserving self-paced contrastive learning model enhances discriminability for better cross-modal segmentation.

    More Related Videos

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
    04:48

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    547
    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
    05:56

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

    Published on: April 14, 2023

    2.7K

    Related Experiment Videos

    Last Updated: Oct 7, 2025

    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

    3.0K
    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
    04:48

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    547
    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
    05:56

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

    Published on: April 14, 2023

    2.7K

    Area of Science:

    • Medical image analysis
    • Computer vision
    • Machine learning

    Background:

    • Unsupervised domain adaptation (UDA) commonly aligns feature distributions using adversarial learning, but this global alignment often lacks class-specific discriminability.
    • Existing methods struggle to effectively bridge the gap between source and target domains in medical imaging, particularly for cross-modal segmentation tasks.
    • The category-agnostic nature of conventional UDA limits the exploitation of class-level joint distributions, resulting in less discriminative feature representations.

    Purpose of the Study:

    • To propose a novel margin preserving self-paced contrastive learning (MPSCL) model for effective cross-modal medical image segmentation.
    • To enhance the discriminability of embedded representations by utilizing domain-adaptive category prototypes and a novel margin preserving contrastive loss.
    • To improve category-aware distribution alignment in UDA through self-paced generation of informative pseudo-labels in the target domain.

    Main Methods:

    • Developed a novel margin preserving self-paced contrastive learning (MPSCL) model for unsupervised domain adaptation.
    • Employed domain-adaptive category prototypes to construct positive and negative sample pairs for contrastive learning.
    • Introduced a margin preserving contrastive loss guided by progressively refined semantic prototypes to boost representation discriminability.
    • Generated informative pseudo-labels in a self-paced manner to enhance supervision for contrastive learning and enable category-aware distribution alignment.
    • Learned domain-invariant representations through joint contrastive learning across source and target domains.

    Main Results:

    • The proposed MPSCL model significantly improves semantic segmentation performance in cross-modal medical imaging.
    • MPSCL outperforms existing state-of-the-art methods by a considerable margin on cross-modal cardiac segmentation tasks.
    • The method effectively addresses the limitations of category-agnostic global alignment by focusing on class-level joint distributions.

    Conclusions:

    • MPSCL offers a superior approach to unsupervised domain adaptation for cross-modal medical image segmentation.
    • The proposed techniques, including domain-adaptive prototypes and self-paced pseudo-labeling, enhance feature discriminability and category-aware alignment.
    • This work advances the field by providing a more effective and robust solution for bridging domain gaps in medical image analysis.