Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jul 5, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Unsupervised deep learning with convolutional neural networks for static parallel transmit design: A retrospective

Toygan Kilic^1,2, Patrick Liebig³, Omer Burak Demirel^1,2

¹Electrical and Computer Engineering, University of Minnesota, Minneapolis, Minnesota, USA.

Magnetic Resonance in Medicine

|January 22, 2024

Summary

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

Characterizing Metabolic and Compositional Heterogeneity of Calf Muscle Using CEST MRI at 3 T.

NMR in biomedicine·2026

Same author

An 80-channel receive array for 10.5T neuroimaging: Key considerations for SNR optimization.

bioRxiv : the preprint server for biology·2026

Same author

Real-time AI integration for MR to detect artifacts and guide pulse sequence adaptations.

bioRxiv : the preprint server for biology·2026

Same author

Functional MRI of the Human Hippocampus at 10.5T: Pushing the Boundaries of Spatial Resolution.

bioRxiv : the preprint server for biology·2026

Same author

Bridging the gap with invasive imaging: promises and challenges of a new generation of ultrahigh resolution fMRI.

bioRxiv : the preprint server for biology·2026

Same author

How Much Does Motion Matter? Evaluating the Motion Robustness of pTx Pulses at 7 T.

Magnetic resonance in medicine·2026

This summary is machine-generated.

This study introduces unsupervised deep learning with convolutional neural networks (CNNs) to improve B1+ inhomogeneity in 7T MRI. The novel method outperforms traditional solutions in speed and robustness for parallel transmit (pTx) applications.

Area of Science:

Magnetic Resonance Imaging (MRI)
Artificial Intelligence in Medical Imaging
Biophysics

Background:

B1+ inhomogeneity is a significant challenge in high-field (7T) MRI, particularly with multi-channel transmit arrays.
Existing deep learning methods for parallel transmit (pTx) pulse design often rely on supervised training and do not effectively utilize multi-channel B1+ maps.
Convolutional Neural Networks (CNNs) offer potential for processing complex spatial data like B1+ maps.

Purpose of the Study:

To develop and evaluate an unsupervised deep learning approach using CNNs to mitigate B1+ inhomogeneity at 7T for multi-channel transmit arrays.
To enable the use of CNNs with multi-channel B1+ maps through an unsupervised training strategy.
To eliminate the need for calculating reference transmit RF weights by employing a physics-driven loss function.

Keywords:

7 T RF inhomogeneity mitigation convolutional neural networks deep learning parallel excitation unsupervised learning

More Related Videos

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

Published on: May 10, 2024

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

Related Experiment Videos

Last Updated: Jul 5, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

Published on: May 10, 2024

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

Main Methods:

Concatenated multi-channel B1+ maps along the spatial dimension for shift-equivariant processing suitable for CNNs.
Implemented unsupervised training utilizing a physics-driven loss function that minimizes Bloch simulation discrepancies.
Trained the model on a dataset of 3824 2D sagittal multi-channel B1+ maps from 143 healthy subjects acquired at 7T.
Compared the proposed method against the unregularized magnitude least-squares (MLS) solution for static pTx design.

Main Results:

The unsupervised CNN method demonstrated superior performance over unregularized MLS in reducing Root Mean Square (RMS) error and coefficient-of-variation.
The proposed method achieved comparable energy consumption to the MLS solution.
Unlike the unregularized MLS solution, the proposed method avoided local phase singularities, preventing holes in the resulting magnetization.

Conclusions:

Unsupervised deep learning with CNNs offers a more effective and robust solution for static parallel transmit (pTx) compared to unregularized MLS.
The developed method improves speed and robustness in addressing B1+ inhomogeneity at 7T.
This approach facilitates the application of CNNs to multi-channel B1+ mapping in a computationally efficient manner.