Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques

IntroductionA 3T MRI scanner delivers enhanced image quality and SNR, minimizing artifacts to provide superior high-resolution brain images compared to a 1.5T MRI. Thus, making it vitally important for diagnosing complex neurological conditions. However, its higher cost of acquisition and operation,...

Full description

Saved in:
Bibliographic Details
Main Authors: Sk Rahatul Jannat, Kirsten Lynch, Maryam Fotouhi, Steve Cen, Jeiran Choupan, Nasim Sheikh-Bahaei, Gaurav Pandey, Bino A. Varghese
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-06-01
Series:Frontiers in Human Neuroscience
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fnhum.2025.1532395/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850218976605372416
author Sk Rahatul Jannat
Kirsten Lynch
Maryam Fotouhi
Steve Cen
Steve Cen
Jeiran Choupan
Nasim Sheikh-Bahaei
Nasim Sheikh-Bahaei
Gaurav Pandey
Bino A. Varghese
author_facet Sk Rahatul Jannat
Kirsten Lynch
Maryam Fotouhi
Steve Cen
Steve Cen
Jeiran Choupan
Nasim Sheikh-Bahaei
Nasim Sheikh-Bahaei
Gaurav Pandey
Bino A. Varghese
author_sort Sk Rahatul Jannat
collection DOAJ
description IntroductionA 3T MRI scanner delivers enhanced image quality and SNR, minimizing artifacts to provide superior high-resolution brain images compared to a 1.5T MRI. Thus, making it vitally important for diagnosing complex neurological conditions. However, its higher cost of acquisition and operation, increased sensitivity to image distortions, greater noise levels, and limited accessibility in many healthcare settings present notable challenges. These factors impact heterogeneity in MRI neuroimaging data on account of the uneven distribution of 1.5T and 3T MRI scanners across healthcare institutions.MethodsIn our study, we investigated the efficacy of three deep learning-based super-resolution techniques to enhance 1.5T MRI images, aiming to achieve quality analogous to 3T scans. These synthetic and “upgraded” 1.5T images were compared and assessed against their 3T counterparts using a range of image quality assessment metrics. Specifically, we employed metrics such as the Structural Similarity Index Measure (SSIM), Peak Signal-to-Noise Ratio (PSNR), Learned Perceptual Image Patch Similarity (LPIPS), and Intensity Differences in Pixels (IDP) to evaluate the similitude and visual quality of the enhanced images.ResultsAccording to our experimental results it has been exhibited that among the three evaluated deep learning-based super-resolution techniques, the Transformer Enhanced Generative Adversarial Network (TCGAN) significantly outperformed the others. To reduce pixel differences, enhance image sharpness, and preserve essential anatomical details TCGAN performed efficaciously.DiscussionThis approach presents TCGAN offers a cost-effective and widely accessible alternative for generating high-quality images without the need for expensive, high-field MRI scans and leads to inconsistencies and complicate data comparison and harmonization challenges across studies utilizing various scanners.
format Article
id doaj-art-e3007ec2b2d249659a4763ef9a470d74
institution OA Journals
issn 1662-5161
language English
publishDate 2025-06-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Human Neuroscience
spelling doaj-art-e3007ec2b2d249659a4763ef9a470d742025-08-20T02:07:31ZengFrontiers Media S.A.Frontiers in Human Neuroscience1662-51612025-06-011910.3389/fnhum.2025.15323951532395Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniquesSk Rahatul Jannat0Kirsten Lynch1Maryam Fotouhi2Steve Cen3Steve Cen4Jeiran Choupan5Nasim Sheikh-Bahaei6Nasim Sheikh-Bahaei7Gaurav Pandey8Bino A. Varghese9Department of Radiology, University of Southern California, Los Angeles, CA, United StatesDepartment of Neurology, University of Southern California, Los Angeles, CA, United StatesDepartment of Radiology, University of Southern California, Los Angeles, CA, United StatesDepartment of Radiology, University of Southern California, Los Angeles, CA, United StatesDepartment of Neurology, University of Southern California, Los Angeles, CA, United StatesDepartment of Neurology, University of Southern California, Los Angeles, CA, United StatesDepartment of Radiology, University of Southern California, Los Angeles, CA, United StatesDepartment of Neurology, University of Southern California, Los Angeles, CA, United StatesDepartment of Genetics & Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, NY, United StatesDepartment of Radiology, University of Southern California, Los Angeles, CA, United StatesIntroductionA 3T MRI scanner delivers enhanced image quality and SNR, minimizing artifacts to provide superior high-resolution brain images compared to a 1.5T MRI. Thus, making it vitally important for diagnosing complex neurological conditions. However, its higher cost of acquisition and operation, increased sensitivity to image distortions, greater noise levels, and limited accessibility in many healthcare settings present notable challenges. These factors impact heterogeneity in MRI neuroimaging data on account of the uneven distribution of 1.5T and 3T MRI scanners across healthcare institutions.MethodsIn our study, we investigated the efficacy of three deep learning-based super-resolution techniques to enhance 1.5T MRI images, aiming to achieve quality analogous to 3T scans. These synthetic and “upgraded” 1.5T images were compared and assessed against their 3T counterparts using a range of image quality assessment metrics. Specifically, we employed metrics such as the Structural Similarity Index Measure (SSIM), Peak Signal-to-Noise Ratio (PSNR), Learned Perceptual Image Patch Similarity (LPIPS), and Intensity Differences in Pixels (IDP) to evaluate the similitude and visual quality of the enhanced images.ResultsAccording to our experimental results it has been exhibited that among the three evaluated deep learning-based super-resolution techniques, the Transformer Enhanced Generative Adversarial Network (TCGAN) significantly outperformed the others. To reduce pixel differences, enhance image sharpness, and preserve essential anatomical details TCGAN performed efficaciously.DiscussionThis approach presents TCGAN offers a cost-effective and widely accessible alternative for generating high-quality images without the need for expensive, high-field MRI scans and leads to inconsistencies and complicate data comparison and harmonization challenges across studies utilizing various scanners.https://www.frontiersin.org/articles/10.3389/fnhum.2025.1532395/fullimage qualitysuper resolutionT1 weightedImage harmonizationtransformer enhanced GAN
spellingShingle Sk Rahatul Jannat
Kirsten Lynch
Maryam Fotouhi
Steve Cen
Steve Cen
Jeiran Choupan
Nasim Sheikh-Bahaei
Nasim Sheikh-Bahaei
Gaurav Pandey
Bino A. Varghese
Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques
Frontiers in Human Neuroscience
image quality
super resolution
T1 weighted
Image harmonization
transformer enhanced GAN
title Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques
title_full Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques
title_fullStr Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques
title_full_unstemmed Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques
title_short Advancing 1.5T MR imaging: toward achieving 3T quality through deep learning super-resolution techniques
title_sort advancing 1 5t mr imaging toward achieving 3t quality through deep learning super resolution techniques
topic image quality
super resolution
T1 weighted
Image harmonization
transformer enhanced GAN
url https://www.frontiersin.org/articles/10.3389/fnhum.2025.1532395/full
work_keys_str_mv AT skrahatuljannat advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT kirstenlynch advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT maryamfotouhi advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT stevecen advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT stevecen advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT jeiranchoupan advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT nasimsheikhbahaei advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT nasimsheikhbahaei advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT gauravpandey advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques
AT binoavarghese advancing15tmrimagingtowardachieving3tqualitythroughdeeplearningsuperresolutiontechniques