QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field

QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field

Quantitative susceptibility mapping (QSM) is a powerful MRI technique that has shown great potential in quantifying tissue susceptibility in numerous neurological disorders. However, the intrinsic ill-posed dipole inversion problem greatly affects the accuracy of the susceptibility map. We propose Q...

Full description

Bibliographic Details
Main Authors: Yicheng Chen, Angela Jakary, Sivakami Avadiappan, Christopher P. Hess, Janine M. Lupo
Format: Article
Language:English
Published: Elsevier 2020-02-01
Series:NeuroImage
Subjects:
Magnetic resonance imaging
Quantitative susceptibility mapping
Dipole field inversion
Deep convolutional neural networks
Generative adversarial networks
Cerebral microbleeds
Online Access:http://www.sciencedirect.com/science/article/pii/S1053811919309802
id doaj-03d69dbeae974f28afcb76c330d81a28
record_format Article
spelling doaj-03d69dbeae974f28afcb76c330d81a282020-11-25T03:51:38ZengElsevierNeuroImage1095-95722020-02-01207116389QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive fieldYicheng Chen0Angela Jakary1Sivakami Avadiappan2Christopher P. Hess3Janine M. Lupo4From the UCSF/UC Berkeley Graduate Program in Bioengineering, University of California, San Francisco and Berkeley, CA, USA; From the Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, USAFrom the Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, USAFrom the Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, USAFrom the Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, USAFrom the UCSF/UC Berkeley Graduate Program in Bioengineering, University of California, San Francisco and Berkeley, CA, USA; From the Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, USA; Corresponding author. Byers Hall UCSF, Box 2532, 1700 4th Street, Suite 303D, San Francisco, CA, 94158-2330, USA.Quantitative susceptibility mapping (QSM) is a powerful MRI technique that has shown great potential in quantifying tissue susceptibility in numerous neurological disorders. However, the intrinsic ill-posed dipole inversion problem greatly affects the accuracy of the susceptibility map. We propose QSMGAN: a 3D deep convolutional neural network approach based on a 3D U-Net architecture with increased receptive field of the input phase compared to the output and further refined the network using the WGAN with gradient penalty training strategy. Our method generates accurate QSM maps from single orientation phase maps efficiently and performs significantly better than traditional non-learning-based dipole inversion algorithms. The generalization capability was verified by applying the algorithm to an unseen pathology--brain tumor patients with radiation-induced cerebral microbleeds.http://www.sciencedirect.com/science/article/pii/S1053811919309802Magnetic resonance imagingQuantitative susceptibility mappingDipole field inversionDeep convolutional neural networksGenerative adversarial networksCerebral microbleeds
collection DOAJ
language English
format Article
sources DOAJ
author Yicheng Chen
Angela Jakary
Sivakami Avadiappan
Christopher P. Hess
Janine M. Lupo
spellingShingle Yicheng Chen
Angela Jakary
Sivakami Avadiappan
Christopher P. Hess
Janine M. Lupo
QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field
NeuroImage
Magnetic resonance imaging
Quantitative susceptibility mapping
Dipole field inversion
Deep convolutional neural networks
Generative adversarial networks
Cerebral microbleeds
author_facet Yicheng Chen
Angela Jakary
Sivakami Avadiappan
Christopher P. Hess
Janine M. Lupo
author_sort Yicheng Chen
title QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field
title_short QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field
title_full QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field
title_fullStr QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field
title_full_unstemmed QSMGAN: Improved Quantitative Susceptibility Mapping using 3D Generative Adversarial Networks with increased receptive field
title_sort qsmgan: improved quantitative susceptibility mapping using 3d generative adversarial networks with increased receptive field
publisher Elsevier
series NeuroImage
issn 1095-9572
publishDate 2020-02-01
description Quantitative susceptibility mapping (QSM) is a powerful MRI technique that has shown great potential in quantifying tissue susceptibility in numerous neurological disorders. However, the intrinsic ill-posed dipole inversion problem greatly affects the accuracy of the susceptibility map. We propose QSMGAN: a 3D deep convolutional neural network approach based on a 3D U-Net architecture with increased receptive field of the input phase compared to the output and further refined the network using the WGAN with gradient penalty training strategy. Our method generates accurate QSM maps from single orientation phase maps efficiently and performs significantly better than traditional non-learning-based dipole inversion algorithms. The generalization capability was verified by applying the algorithm to an unseen pathology--brain tumor patients with radiation-induced cerebral microbleeds.
topic Magnetic resonance imaging
Quantitative susceptibility mapping
Dipole field inversion
Deep convolutional neural networks
Generative adversarial networks
Cerebral microbleeds
url http://www.sciencedirect.com/science/article/pii/S1053811919309802
work_keys_str_mv AT yichengchen qsmganimprovedquantitativesusceptibilitymappingusing3dgenerativeadversarialnetworkswithincreasedreceptivefield
AT angelajakary qsmganimprovedquantitativesusceptibilitymappingusing3dgenerativeadversarialnetworkswithincreasedreceptivefield
AT sivakamiavadiappan qsmganimprovedquantitativesusceptibilitymappingusing3dgenerativeadversarialnetworkswithincreasedreceptivefield
AT christopherphess qsmganimprovedquantitativesusceptibilitymappingusing3dgenerativeadversarialnetworkswithincreasedreceptivefield
AT janinemlupo qsmganimprovedquantitativesusceptibilitymappingusing3dgenerativeadversarialnetworkswithincreasedreceptivefield
_version_ 1724486459758477312

Similar Items