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

Betulinic acid ameliorates experimental diabetic-induced renal inflammation and fibrosis via inhibiting the activation of NF-κB signaling pathway.

Molecular and cellular endocrinology·2016
Same author

Primary pulmonary T-cell lymphoma mimicking pneumonia: A case report and literature review.

Experimental and therapeutic medicine·2016
Same author

Investigating polymorphisms by bioinformatics is a potential cost-effective method to screen for germline mutations in Chinese familial adenomatous polyposis patients.

Oncology letters·2016
Same author

Discovery and characterization of a novel potent type II native and mutant BCR-ABL inhibitor (CHMFL-074) for Chronic Myeloid Leukemia (CML).

Oncotarget·2016
Same author

Temporal Patterns in Bacterioplankton Community Composition in Three Reservoirs of Similar Trophic Status in Shenzhen, China.

International journal of environmental research and public health·2016
Same author

Expanding CRISPR/Cas9 Genome Editing Capacity in Zebrafish Using SaCas9.

G3 (Bethesda, Md.)·2016

Related Experiment Video

Updated: Sep 12, 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

GM-ABS: Promptable Generalist Model Drives Active Barely Supervised Training in Specialist Model for 3D Medical Image

Zhe Xu, Cheng Chen, Donghuan Lu

    IEEE Transactions on Medical Imaging
    |August 7, 2025
    PubMed
    Summary

    Generalist models like SAM enhance semi-supervised learning (SSL) for 3D medical image segmentation. Our GM-ABS approach uses specialist-generalist collaboration and active learning for minimal annotation, improving segmentation accuracy with limited data.

    More Related Videos

    Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions
    06:18

    Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions

    Published on: April 5, 2024

    1.2K
    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 12, 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: Segmentation and VR for Advanced Neurovascular Interventions
    06:18

    Author Spotlight: Segmentation and VR for Advanced Neurovascular Interventions

    Published on: April 5, 2024

    1.2K
    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
    • Artificial Intelligence
    • Computer Vision

    Background:

    • Semi-supervised learning (SSL) reduces the need for extensive manual labeling in 3D medical image segmentation.
    • Foundational generalist models, such as the Segment Anything Model (SAM), offer zero-shot segmentation capabilities but often lag behind specialists in performance.
    • These generalist models present an opportunity to enhance SSL by providing a data-centric perspective, particularly for pseudo and expert labeling strategies.

    Purpose of the Study:

    • To propose the Generalist Model-driven Active Barely Supervised (GM-ABS) learning paradigm.
    • To develop specialized 3D segmentation models using extremely limited annotation budgets (e.g., three slices per scan).
    • To revolutionize pseudo and expert labeling strategies within SSL for 3D medical imaging.

    Main Methods:

    • GM-ABS builds upon a mean-teacher SSL framework, incorporating two key data-centric designs.
    • Specialist-generalist collaboration: An in-training specialist uses class-specific prompts to interact with a frozen generalist model (SAM) across multiple views for noisy label augmentation.
    • Expert-model collaboration: Active cross-labeling with minimal expert effort, guided by a human-in-the-loop approach, progressively enhances specialist supervision.

    Main Results:

    • GM-ABS demonstrated promising performance on three benchmark datasets.
    • The proposed method achieved superior results compared to recent SSL approaches under extremely constrained labeling resources.
    • Specialist-generalist collaboration effectively leveraged prompts for noisy label augmentation and noise-tolerant assimilation.

    Conclusions:

    • GM-ABS offers an effective solution for training specialized 3D medical segmentation models with minimal annotations.
    • The paradigm highlights the potential of leveraging generalist models and active learning for data-centric SSL in medical imaging.
    • GM-ABS significantly enhances data pool utilization and segmentation accuracy under severe annotation limitations.