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

GPR177 in A-fiber sensory neurons drives diabetic neuropathic pain via WNT-mediated TRPV1 activation.

Science translational medicine·2022
Same author

Lesion-attention pyramid network for diabetic retinopathy grading.

Artificial intelligence in medicine·2022
Same author

Development and Validation of a Nomogram to Predict Cancer-Specific Survival for Middle-Aged Patients With Early-Stage Hepatocellular Carcinoma.

Frontiers in public health·2022
Same author

Expanding the DNA-encoded library toolbox: identifying small molecules targeting RNA.

Nucleic acids research·2022
Same author

Depression duration and risk of incident cardiovascular disease: A population-based six-year cohort study.

Journal of affective disorders·2022
Same author

Adverse Events Reporting of Clinical Trials in Exercise Oncology Research (ADVANCE): Protocol for a Scoping Review.

Frontiers in oncology·2022

Related Experiment Video

Updated: Aug 3, 2025

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

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

470

SDMT: Spatial Dependence Multi-Task Transformer Network for 3D Knee MRI Segmentation and Landmark Localization.

Xiang Li, Songcen Lv, Minglei Li

    IEEE Transactions on Medical Imaging
    |April 7, 2023
    PubMed
    Summary
    This summary is machine-generated.

    This study introduces a Spatial Dependence Multi-task Transformer (SDMT) for 3D knee MRI analysis. The novel network effectively performs both segmentation and landmark localization, improving diagnostic accuracy for knee conditions.

    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

    2.8K
    A Pipeline for 3D Multimodality Image Integration and Computer-assisted Planning in Epilepsy Surgery
    09:41

    A Pipeline for 3D Multimodality Image Integration and Computer-assisted Planning in Epilepsy Surgery

    Published on: May 20, 2016

    12.4K

    Related Experiment Videos

    Last Updated: Aug 3, 2025

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

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    470
    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

    2.8K
    A Pipeline for 3D Multimodality Image Integration and Computer-assisted Planning in Epilepsy Surgery
    09:41

    A Pipeline for 3D Multimodality Image Integration and Computer-assisted Planning in Epilepsy Surgery

    Published on: May 20, 2016

    12.4K

    Area of Science:

    • Medical Imaging
    • Artificial Intelligence
    • Biomedical Engineering

    Background:

    • 3D knee MRI analysis involves complex segmentation and landmark localization tasks crucial for diagnosing and treating knee diseases.
    • Current deep learning approaches often use single-task Convolutional Neural Network (CNN) models, which present challenges due to the intricate knee anatomy and clinical usability limitations.
    • Developing independent models for each task increases complexity for surgeons in clinical settings.

    Purpose of the Study:

    • To propose a novel Spatial Dependence Multi-task Transformer (SDMT) network for simultaneous 3D knee MRI segmentation and landmark localization.
    • To leverage the spatial dependencies between segmentation and landmark localization to mutually enhance performance.
    • To provide a unified and clinically applicable deep learning solution for comprehensive knee MRI analysis.

    Main Methods:

    • A shared encoder is employed for efficient feature extraction from 3D knee MRI data.
    • The SDMT network incorporates spatial encoding and a task-hybridized multi-head attention mechanism with inter-task and intra-task attention heads.
    • A dynamic weight multi-task loss function is designed to effectively balance the training of both segmentation and landmark localization tasks.

    Main Results:

    • The proposed SDMT method achieved a Dice score of 83.91% for the knee segmentation task.
    • For landmark localization, the method obtained a Mean Radial Error (MRE) of 2.12 mm.
    • Performance was competitive and demonstrated superiority over existing state-of-the-art single-task methods.

    Conclusions:

    • The SDMT network offers a robust and effective multi-task learning framework for 3D knee MRI.
    • Simultaneous segmentation and landmark localization using SDMT improves accuracy and clinical utility.
    • This approach provides a promising direction for advancing automated knee MRI analysis in clinical practice.