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

Anthropogenic gadolinium contamination in tap water across the Paris megacity: Rare earth elements trace sources and mixing.

Environmental pollution (Barking, Essex : 1987)·2026
Same author

AI-driven volumetric approach for automatic chemotherapy response assessment in colorectal liver metastases.

European radiology·2026
Same author

Rare earth element abundances and gadolinium contamination in tap water worldwide.

Chemosphere·2025
Same author

Bias, accuracy, and trust: no GenAI in peer reviewing.

Journal of neuroradiology = Journal de neuroradiologie·2025
Same author

Exploring learning transferability in deep segmentation of colorectal cancer liver metastases.

Computers in biology and medicine·2025
Same author

Unveiling rare earth elements in beers:evidence for gadolinium contamination.

Food chemistry·2025

Related Experiment Video

Updated: Jul 25, 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

2.8K

Cross-dimensional transfer learning in medical image segmentation with deep learning.

Hicham Messaoudi1, Ahror Belaid2, Douraied Ben Salem3

  • 1Laboratory of Medical Informatics (LIMED), Faculty of Technology, University of Bejaia, 06000 Bejaia, Algeria.

Medical Image Analysis
|June 29, 2023
PubMed
Summary

This study presents a novel method for medical image segmentation by transferring knowledge from 2D natural image networks. The approach significantly improves segmentation accuracy for various medical imaging modalities.

Keywords:
Convolutional neural networksCross-dimensional transferMedical image segmentationTransfer learning

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

444
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

8.9K

Related Experiment Videos

Last Updated: Jul 25, 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

2.8K
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

444
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

8.9K

Area of Science:

  • Computer Vision
  • Medical Image Analysis
  • Deep Learning

Background:

  • Convolutional neural networks (CNNs) excel in natural image analysis but face challenges in medical imaging due to limited annotated data and acquisition constraints.
  • Medical imaging data often involves high dimensionality, posing challenges for existing 2D segmentation models.
  • State-of-the-art performance in medical image segmentation is hindered by data scarcity and inherent imaging complexities.

Purpose of the Study:

  • To develop an efficient method for transferring knowledge from 2D natural image classification networks to 2D and 3D medical image segmentation tasks.
  • To introduce novel network architectures that leverage pre-trained 2D encoders and facilitate dimensional expansion for enhanced segmentation.
  • To validate the proposed methods on diverse medical imaging datasets and benchmark challenges.

Main Methods:

  • Designed novel U-Net architectures incorporating a 2D pre-trained encoder for weight transfer.
  • Developed a dimensional transfer approach to expand 2D segmentation networks into higher dimensions.
  • Tested networks on multi-modal datasets including MRI, CT, and ultrasound images from challenges like CAMUS, CHAOS, and BraTS.

Main Results:

  • Achieved state-of-the-art results on the CAMUS challenge for echocardiographic data segmentation.
  • Outperformed other 2D-based methods on the CHAOS challenge for abdominal CT and MRI segmentation.
  • Attained high Dice scores on the BraTS 2022 competition for brain tumor segmentation using 3D networks.

Conclusions:

  • The proposed weight and dimensional transfer methods effectively adapt 2D CNNs for accurate multi-dimensional medical image segmentation.
  • The approach demonstrates significant improvements in segmentation accuracy across various medical imaging modalities and tasks.
  • This work offers a promising solution to overcome data limitations in medical image analysis, enhancing diagnostic capabilities.