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: Dec 7, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Transfer learning in deep neural network based under-sampled MR image reconstruction.

Madiha Arshad¹, Mahmood Qureshi¹, Omair Inam¹

¹Medical Image Processing Research Group (MIPRG), Department of Electrical and Computer Engineering, COMSATS University Islamabad, Pakistan.

Magnetic Resonance Imaging

|September 27, 2020

Summary

This summary is machine-generated.

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

Enhanced respiratory motion compensation in free-breathing dynamic contrast-enhanced MRI with GROG-facilitated bunch phase encoding and Golden angle radial sampling.

Magnetic resonance imaging·2026

Same author

SC-GROG followed by L+S reconstruction with multiple sparsity constraints for accelerated Golden-angle-radial DCE-MRI.

PloS one·2025

Same author

Cardiac MR image reconstruction using cascaded hybrid dual domain deep learning framework.

PloS one·2025

Same author

Brain tumor detection and segmentation using deep learning.

Magma (New York, N.Y.)·2024

Same author

The Power of Advocacy in Global Neurosurgery.

Neurosurgery·2024

Same author

Global Neurosurgery: An Overview.

Neurosurgery·2024

Same journal

Repeatability of an MRI protocol for generating habitats based on cellularity, perfusion, and hypoxia in a murine model of glioma.

Magnetic resonance imaging·2026

Same journal

Association between CMR-derived pulmonary artery pulse wave velocity and pulmonary risk factors: the multi-ethnic study of atherosclerosis COPD study.

Magnetic resonance imaging·2026

Same journal

Systematic comparison of MPRAGE and BRAVO T1-weighted MRI pulse sequences and brain morphometry in high-risk young adults.

Magnetic resonance imaging·2026

Same journal

Foot dynamic contrast-enhanced MRI for assessing microcirculatory changes after endovascular therapy in peripheral artery disease: A prospective pilot study.

Magnetic resonance imaging·2026

Same journal

Reconstruction of MRI from undersampled k-spaces of double-contrast volume acquisitions using deep neural networks.

Magnetic resonance imaging·2026

Same journal

Radiofrequency-induced heating safety of brain MRI scans at 7 T in the presence of a shoulder implant.

Magnetic resonance imaging·2026

See all related articles

Transfer learning with U-Net overcomes data scarcity in Magnetic Resonance Imaging (MRI) reconstruction. This approach enables effective deep learning model adaptation for diverse MRI data without extensive retraining.

Area of Science:

Medical Imaging
Artificial Intelligence
Deep Learning

Background:

Deep learning for under-sampled MRI reconstruction requires large datasets and robust generalization.
Retraining neural networks for new MRI protocols is data-intensive, costly, and time-consuming.
Existing methods struggle with data scarcity and generalization across different MRI acquisition parameters.

Purpose of the Study:

To propose and evaluate a transfer learning approach for U-Net based MR image reconstruction.
To address data scarcity and improve generalization of deep learning models in MRI.
To enable efficient reconstruction of MR images across varying magnetic field strengths, anatomies, and acceleration factors.

Main Methods:

Utilized a pre-trained U-Net model, initially trained on human brain images from a 1.5T scanner.

Keywords:

CG-SENSE Data scarcity Deep learning Generalization capabilities MRI Transfer learning

Related Experiment Videos

Last Updated: Dec 7, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Assessed the generalization of the pre-trained U-Net on MR images from different field strengths, anatomies, and undersampling factors.

Applied end-to-end fine-tuning of the pre-trained U-Net for reconstruction tasks involving diverse MR image datasets.

Main Results:

Demonstrated successful MR image reconstruction using the proposed transfer learning method.
Evaluated reconstruction quality using Structural Similarity Index (SSIM), Root Mean Square Error (RMSE), and Peak Signal-to-Noise Ratio (PSNR).
Achieved effective reconstruction across varied magnetic field strengths, anatomical regions, and undersampling levels.

Conclusions:

Transfer learning, specifically end-to-end fine-tuning, is a viable solution for data scarcity in deep learning-based MRI reconstruction.
The proposed method enhances the generalization capabilities of U-Net models for diverse MRI applications.
This approach reduces the need for extensive data acquisition and retraining, making MRI reconstruction more efficient.