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

Cartesian MPnRAGE for Efficient Simultaneous Multi-Contrast and Quantitative Relaxometry Imaging.

Magnetic resonance in medicine·2026
Same author

Longitudinal <sup>1</sup>H and <sup>129</sup>Xe Lung MRI in Patients With Post-COVID Residual Lung Abnormalities.

Journal of magnetic resonance imaging : JMRI·2026
Same author

Comparison of Retrospective Motion Compensation Techniques for Pulmonary Dynamic Ultrashort Time to Echo MRI in Suspected Idiopathic Pulmonary Fibrosis.

Journal of magnetic resonance imaging : JMRI·2026
Same author

Effects and Correction of Patient Bulk Motion in Cranial DENSE MRI.

Magnetic resonance in medicine·2026
Same author

Jointly Learned 3D Non-Cartesian Sampling With Wave Encoding and Reconstruction for Neurovascular Phase Contrast MRI.

Magnetic resonance in medicine·2025
Same author

Quantifying physiological variability and improving reproducibility in 4D-flow MRI cerebrovascular measurements with self-supervised deep learning.

Magnetic resonance in medicine·2025

Related Experiment Video

Updated: Aug 11, 2025

Author Spotlight: Optimized Lung MRI Protocol with Computationally Efficient Reconstruction Methods
05:07

Author Spotlight: Optimized Lung MRI Protocol with Computationally Efficient Reconstruction Methods

Published on: September 6, 2024

444

Motion compensated self supervised deep learning for highly accelerated 3D ultrashort Echo time pulmonary MRI.

Zachary Miller1, Kevin M Johnson2,3

  • 1Department of Biomedical Engineering, University of Wisconsin, Madison, Wisconsin, USA.

Magnetic Resonance in Medicine
|February 6, 2023
PubMed
Summary

Motion compensated, self-supervised, model-based deep learning (MBDL) reconstructs high-quality 3D pulmonary UTE images from free-breathing scans. This advanced XD-MBDL method improves image quality and reduces reconstruction time compared to existing techniques.

Keywords:
image reconstructionmodel based deep learningmotion-compensation

More Related Videos

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging
10:44

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Published on: June 21, 2024

571
Author Spotlight: Standardization and Best Practices for Advancing Lung Imaging Using 129Xe MRI
09:08

Author Spotlight: Standardization and Best Practices for Advancing Lung Imaging Using 129Xe MRI

Published on: November 21, 2023

936

Related Experiment Videos

Last Updated: Aug 11, 2025

Author Spotlight: Optimized Lung MRI Protocol with Computationally Efficient Reconstruction Methods
05:07

Author Spotlight: Optimized Lung MRI Protocol with Computationally Efficient Reconstruction Methods

Published on: September 6, 2024

444
Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging
10:44

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Published on: June 21, 2024

571
Author Spotlight: Standardization and Best Practices for Advancing Lung Imaging Using 129Xe MRI
09:08

Author Spotlight: Standardization and Best Practices for Advancing Lung Imaging Using 129Xe MRI

Published on: November 21, 2023

936

Area of Science:

  • Medical Imaging
  • Deep Learning
  • Pulmonary Imaging

Background:

  • Free-breathing 3D pulmonary UTE acquisitions present reconstruction challenges due to respiratory motion.
  • Current methods struggle to effectively combine dynamic respiratory states for high-quality single-image reconstruction.
  • Motion compensation is crucial for improving image quality in dynamic pulmonary imaging.

Purpose of the Study:

  • To investigate motion compensated, self-supervised, model-based deep learning (MBDL) for reconstructing free-breathing 3D pulmonary UTE acquisitions.
  • To develop and evaluate an eXtra dimension MBDL (XD-MBDL) architecture that integrates respiratory states and motion correction.
  • To compare the performance of XD-MBDL against existing reconstruction techniques.

Main Methods:

  • Developed a self-supervised XD-MBDL architecture combining respiratory states for single 3D image reconstruction.
  • Incorporated non-rigid motion fields by estimating them from lower-resolution XD-GRASP reconstructions.
  • Evaluated motion-compensated XD-MBDL on lung UTE datasets with and without contrast, comparing it to constrained reconstructions and other MBDL variants.

Main Results:

  • XD-MBDL demonstrated improved apparent SNR (aSNR) and contrast-to-noise ratio (CNR) compared to non-motion-compensated MBDL, XD-GRASP, and iMoCo.
  • Visual assessment confirmed superior image quality with XD-MBDL.
  • XD-MBDL achieved reduced reconstruction times relative to XD-GRASP and iMoCo.

Conclusions:

  • A novel method was developed enabling self-supervised MBDL to integrate multiple respiratory states into a single image reconstruction.
  • Graphics processing unit (GPU)-based image registration enhanced reconstruction quality when combined with XD-MBDL.
  • The motion-compensated XD-MBDL approach shows promise for reconstructing desired respiratory phases from free-breathing 3D pulmonary UTE data.