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

Needle immersed vitrification can lower the concentration of cryoprotectant in human ovarian tissue cryopreservation.

Fertility and sterility·2010
Same author

Maternal control of early mouse development.

Development (Cambridge, England)·2010
Same author

Characterization of EndoTT, a novel single-stranded DNA-specific endonuclease from Thermoanaerobacter tengcongensis.

Nucleic acids research·2010
Same author

Association study between three polymorphisms and myocardial infarction and ischemic stroke in Chinese Han population.

Thrombosis research·2010
Same author

Arabidopsis IWS1 interacts with transcription factor BES1 and is involved in plant steroid hormone brassinosteroid regulated gene expression.

Proceedings of the National Academy of Sciences of the United States of America·2010
Same author

Effect of isoflavone extracts from glycine max on human endothelial cell damage and on nitric oxide production.

Menopause (New York, N.Y.)·2010

Related Experiment Video

Updated: Oct 8, 2025

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

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

548

TransDIR: Deformable imaging registration network based on transformer to improve the feature extraction ability.

Tiejun Yang1,2,3, Xinhao Bai4, Xiaojuan Cui4

  • 1Key Laboratory of Grain Information Processing and Control (HAUT), Ministry of Education, Zhengzhou, China.

Medical Physics
|December 24, 2021
PubMed
Summary

A novel Transformer-based network, TransDIR, enhances 3D deformable image registration by effectively capturing global and local features. This approach improves accuracy and smoothness in medical image analysis, outperforming existing methods.

Keywords:
global feature extractionregistrationtransformerzero padding

More Related Videos

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.0K
Multi-modal Pulmonary Imaging: Using Complementary Information from CT and Hyperpolarized 129Xe MRI to Evaluate Lung Structure-Function
02:09

Multi-modal Pulmonary Imaging: Using Complementary Information from CT and Hyperpolarized 129Xe MRI to Evaluate Lung Structure-Function

Published on: April 12, 2024

724

Related Experiment Videos

Last Updated: Oct 8, 2025

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

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

548
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.0K
Multi-modal Pulmonary Imaging: Using Complementary Information from CT and Hyperpolarized 129Xe MRI to Evaluate Lung Structure-Function
02:09

Multi-modal Pulmonary Imaging: Using Complementary Information from CT and Hyperpolarized 129Xe MRI to Evaluate Lung Structure-Function

Published on: April 12, 2024

724

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Computer Vision

Background:

  • 3D deformable image registration is crucial for medical decision-making.
  • Convolutional neural network-based methods struggle with global feature extraction, limiting registration performance.
  • Ensuring the smoothness of the displacement vector field (DVF) is a challenge due to folding penalties.

Purpose of the Study:

  • To propose a novel 3D deformable image registration network, TransDIR, leveraging Transformer architecture.
  • To enhance the extraction of both global and local features for improved registration accuracy.
  • To ensure the smoothness of the displacement vector field (DVF) by incorporating a folding penalty.

Main Methods:

  • Developed TransDIR, a 3D deformable image registration network utilizing Transformer.
  • Employed an atrous reduction attention block in the encoder to capture long-distance dependencies and global information.
  • Integrated a zero-padding position encoder for local information capture and an attention-based up-sampling module for Region of Interest (ROI) significance.
  • Incorporated a folding penalty term in the loss function to improve DVF smoothness.

Main Results:

  • TransDIR demonstrated effectiveness on OASIS, LPBA40, MGH10, and MM-WHS datasets.
  • Achieved improved Dice Similarity Coefficient (DSC) scores compared to LapIRN (1.1% on OASIS, 0.9% on LPBA40) and VoxelMorph (2.8% on MM-WHS).
  • Significantly reduced the folding index by hundreds of times on the MM-WHS dataset compared to VoxelMorph.

Conclusions:

  • TransDIR achieves robust medical image registration.
  • The proposed network demonstrates promising generalizability across different datasets.
  • TransDIR outperforms existing methods like LapIRN and VoxelMorph in registration accuracy and DVF smoothness.