Academic Journal
DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction
| العنوان: | DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction |
|---|---|
| المؤلفون: | Yang, G, Yu, S, Dong, H, Slabaugh, G, Dragotti, PL, Ye, X, Liu, F, Arridge, S, Keegan, J, Guo, Y, Firmin, D |
| المصدر: | IEEE Transactions on Medical Imaging (2017) (In press). |
| سنة النشر: | 2017 |
| المجموعة: | University College London: UCL Discovery |
| مصطلحات موضوعية: | Compressed Sensing, Magnetic Resonance Imaging (MRI), Fast MRI, Deep Learning, Generative Adversarial Networks (GAN), De-aliasing, Inverse Problems |
| الوصف: | IEEE Compressed Sensing Magnetic Resonance Imaging (CS-MRI) enables fast acquisition, which is highly desirable for numerous clinical applications. This can not only reduce the scanning cost and ease patient burden, but also potentially reduce motion artefacts and the effect of contrast washout, thus yielding better image quality. Different from parallel imaging based fast MRI, which utilises multiple coils to simultaneously receive MR signals, CS-MRI breaks the Nyquist-Shannon sampling barrier to reconstruct MRI images with much less required raw data. This paper provides a deep learning based strategy for reconstruction of CS-MRI, and bridges a substantial gap between conventional non-learning methods working only on data from a single image, and prior knowledge from large training datasets. In particular, a novel conditional Generative Adversarial Networks-based model (DAGAN) is proposed to reconstruct CS-MRI. In our DAGAN architecture, we have designed a refinement learning method to stabilise our U-Net based generator, which provides an endto-end network to reduce aliasing artefacts. To better preserve texture and edges in the reconstruction, we have coupled the adversarial loss with an innovative content loss. In addition, we incorporate frequency domain information to enforce similarity in both the image and frequency domains. We have performed comprehensive comparison studies with both conventional CSMRI reconstruction methods and newly investigated deep learning approaches. Compared to these methods, our DAGAN method provides superior reconstruction with preserved perceptual image details. Furthermore, each image is reconstructed in about 5 ms, which is suitable for real-time processing. |
| نوع الوثيقة: | article in journal/newspaper |
| وصف الملف: | text |
| اللغة: | English |
| Relation: | https://discovery.ucl.ac.uk/id/eprint/10048154/1/08233175.pdf; https://discovery.ucl.ac.uk/id/eprint/10048154/ |
| الاتاحة: | https://discovery.ucl.ac.uk/id/eprint/10048154/1/08233175.pdf https://discovery.ucl.ac.uk/id/eprint/10048154/ |
| Rights: | open |
| رقم الانضمام: | edsbas.153163EF |
| قاعدة البيانات: | BASE |
| ResultId |
1 |
|---|---|
| Header |
edsbas BASE edsbas.153163EF 871 3 Academic Journal academicJournal 871.106628417969 |
| PLink |
https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&scope=site&db=edsbas&AN=edsbas.153163EF&custid=s6537998&authtype=sso |
| FullText |
Array
(
[Availability] => 0
)
Array ( [0] => Array ( [Url] => https://discovery.ucl.ac.uk/id/eprint/10048154/1/08233175.pdf# [Name] => EDS - BASE [Category] => fullText [Text] => View record in BASE [MouseOverText] => View record in BASE ) ) |
| Items |
Array
(
[Name] => Title
[Label] => Title
[Group] => Ti
[Data] => DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction
)
Array ( [Name] => Author [Label] => Authors [Group] => Au [Data] => <searchLink fieldCode="AR" term="%22Yang%2C+G%22">Yang, G</searchLink><br /><searchLink fieldCode="AR" term="%22Yu%2C+S%22">Yu, S</searchLink><br /><searchLink fieldCode="AR" term="%22Dong%2C+H%22">Dong, H</searchLink><br /><searchLink fieldCode="AR" term="%22Slabaugh%2C+G%22">Slabaugh, G</searchLink><br /><searchLink fieldCode="AR" term="%22Dragotti%2C+PL%22">Dragotti, PL</searchLink><br /><searchLink fieldCode="AR" term="%22Ye%2C+X%22">Ye, X</searchLink><br /><searchLink fieldCode="AR" term="%22Liu%2C+F%22">Liu, F</searchLink><br /><searchLink fieldCode="AR" term="%22Arridge%2C+S%22">Arridge, S</searchLink><br /><searchLink fieldCode="AR" term="%22Keegan%2C+J%22">Keegan, J</searchLink><br /><searchLink fieldCode="AR" term="%22Guo%2C+Y%22">Guo, Y</searchLink><br /><searchLink fieldCode="AR" term="%22Firmin%2C+D%22">Firmin, D</searchLink> ) Array ( [Name] => TitleSource [Label] => Source [Group] => Src [Data] => IEEE Transactions on Medical Imaging (2017) (In press). ) Array ( [Name] => DatePubCY [Label] => Publication Year [Group] => Date [Data] => 2017 ) Array ( [Name] => Subset [Label] => Collection [Group] => HoldingsInfo [Data] => University College London: UCL Discovery ) Array ( [Name] => Subject [Label] => Subject Terms [Group] => Su [Data] => <searchLink fieldCode="DE" term="%22Compressed+Sensing%22">Compressed Sensing</searchLink><br /><searchLink fieldCode="DE" term="%22Magnetic+Resonance+Imaging+%28MRI%29%22">Magnetic Resonance Imaging (MRI)</searchLink><br /><searchLink fieldCode="DE" term="%22Fast+MRI%22">Fast MRI</searchLink><br /><searchLink fieldCode="DE" term="%22Deep+Learning%22">Deep Learning</searchLink><br /><searchLink fieldCode="DE" term="%22Generative+Adversarial+Networks+%28GAN%29%22">Generative Adversarial Networks (GAN)</searchLink><br /><searchLink fieldCode="DE" term="%22De-aliasing%22">De-aliasing</searchLink><br /><searchLink fieldCode="DE" term="%22Inverse+Problems%22">Inverse Problems</searchLink> ) Array ( [Name] => Abstract [Label] => Description [Group] => Ab [Data] => IEEE Compressed Sensing Magnetic Resonance Imaging (CS-MRI) enables fast acquisition, which is highly desirable for numerous clinical applications. This can not only reduce the scanning cost and ease patient burden, but also potentially reduce motion artefacts and the effect of contrast washout, thus yielding better image quality. Different from parallel imaging based fast MRI, which utilises multiple coils to simultaneously receive MR signals, CS-MRI breaks the Nyquist-Shannon sampling barrier to reconstruct MRI images with much less required raw data. This paper provides a deep learning based strategy for reconstruction of CS-MRI, and bridges a substantial gap between conventional non-learning methods working only on data from a single image, and prior knowledge from large training datasets. In particular, a novel conditional Generative Adversarial Networks-based model (DAGAN) is proposed to reconstruct CS-MRI. In our DAGAN architecture, we have designed a refinement learning method to stabilise our U-Net based generator, which provides an endto-end network to reduce aliasing artefacts. To better preserve texture and edges in the reconstruction, we have coupled the adversarial loss with an innovative content loss. In addition, we incorporate frequency domain information to enforce similarity in both the image and frequency domains. We have performed comprehensive comparison studies with both conventional CSMRI reconstruction methods and newly investigated deep learning approaches. Compared to these methods, our DAGAN method provides superior reconstruction with preserved perceptual image details. Furthermore, each image is reconstructed in about 5 ms, which is suitable for real-time processing. ) Array ( [Name] => TypeDocument [Label] => Document Type [Group] => TypDoc [Data] => article in journal/newspaper ) Array ( [Name] => Format [Label] => File Description [Group] => SrcInfo [Data] => text ) Array ( [Name] => Language [Label] => Language [Group] => Lang [Data] => English ) Array ( [Name] => NoteTitleSource [Label] => Relation [Group] => SrcInfo [Data] => https://discovery.ucl.ac.uk/id/eprint/10048154/1/08233175.pdf; https://discovery.ucl.ac.uk/id/eprint/10048154/ ) Array ( [Name] => URL [Label] => Availability [Group] => URL [Data] => https://discovery.ucl.ac.uk/id/eprint/10048154/1/08233175.pdf<br />https://discovery.ucl.ac.uk/id/eprint/10048154/ ) Array ( [Name] => Copyright [Label] => Rights [Group] => Cpyrght [Data] => open ) Array ( [Name] => AN [Label] => Accession Number [Group] => ID [Data] => edsbas.153163EF ) |
| RecordInfo |
Array
(
[BibEntity] => Array
(
[Languages] => Array
(
[0] => Array
(
[Text] => English
)
)
[Subjects] => Array
(
[0] => Array
(
[SubjectFull] => Compressed Sensing
[Type] => general
)
[1] => Array
(
[SubjectFull] => Magnetic Resonance Imaging (MRI)
[Type] => general
)
[2] => Array
(
[SubjectFull] => Fast MRI
[Type] => general
)
[3] => Array
(
[SubjectFull] => Deep Learning
[Type] => general
)
[4] => Array
(
[SubjectFull] => Generative Adversarial Networks (GAN)
[Type] => general
)
[5] => Array
(
[SubjectFull] => De-aliasing
[Type] => general
)
[6] => Array
(
[SubjectFull] => Inverse Problems
[Type] => general
)
)
[Titles] => Array
(
[0] => Array
(
[TitleFull] => DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction
[Type] => main
)
)
)
[BibRelationships] => Array
(
[HasContributorRelationships] => Array
(
[0] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Yang, G
)
)
)
[1] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Yu, S
)
)
)
[2] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Dong, H
)
)
)
[3] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Slabaugh, G
)
)
)
[4] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Dragotti, PL
)
)
)
[5] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Ye, X
)
)
)
[6] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Liu, F
)
)
)
[7] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Arridge, S
)
)
)
[8] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Keegan, J
)
)
)
[9] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Guo, Y
)
)
)
[10] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Firmin, D
)
)
)
)
[IsPartOfRelationships] => Array
(
[0] => Array
(
[BibEntity] => Array
(
[Dates] => Array
(
[0] => Array
(
[D] => 01
[M] => 01
[Type] => published
[Y] => 2017
)
)
[Identifiers] => Array
(
[0] => Array
(
[Type] => issn-locals
[Value] => edsbas
)
[1] => Array
(
[Type] => issn-locals
[Value] => edsbas.oa
)
)
[Titles] => Array
(
[0] => Array
(
[TitleFull] => IEEE Transactions on Medical Imaging (2017) (In press
[Type] => main
)
)
)
)
)
)
)
|
| IllustrationInfo |