Highly undersampled magnetic resonance imaging reconstruction using autoencoding priors

Purpose Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired. Methods A denoising autoencoder (DAE) network is leveraged as an explicit prior to addre...

Full description

Saved in:

Bibliographic Details
Published in:	Magnetic resonance in medicine Vol. 83; no. 1; pp. 322 - 336
Main Authors:	Liu, Qiegen, Yang, Qingxin, Cheng, Huitao, Wang, Shanshan, Zhang, Minghui, Liang, Dong
Format:	Journal Article
Language:	English
Published:	United States Wiley Subscription Services, Inc 01.01.2020
Subjects:	Acceleration Algorithms autoencoding priors Brain - diagnostic imaging Computer Systems Humans Image processing Image Processing, Computer-Assisted - methods Image reconstruction Image restoration Machine learning Magnetic Resonance Imaging multichannel prior Neural Networks, Computer Noise generation Noise reduction proximal gradient descent Sampling Signal-To-Noise Ratio Software Trajectories multichannel prior autoencoding priors magnetic resonance imaging image reconstruction proximal gradient descent
ISSN:	0740-3194, 1522-2594, 1522-2594
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Abstract	Purpose Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired. Methods A denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high‐dimension signals is more effective than that from the low‐dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single‐channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2‐sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent. Results Experimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal‐to‐noise ratio, structural similarity, and high‐frequency error norm. Conclusion A simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state‐of‐the‐art methods.
AbstractList	Purpose Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired. Methods A denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high‐dimension signals is more effective than that from the low‐dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single‐channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2‐sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent. Results Experimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal‐to‐noise ratio, structural similarity, and high‐frequency error norm. Conclusion A simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state‐of‐the‐art methods. PurposeAlthough recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired.MethodsA denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high‐dimension signals is more effective than that from the low‐dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single‐channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2‐sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent.ResultsExperimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal‐to‐noise ratio, structural similarity, and high‐frequency error norm.ConclusionA simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state‐of‐the‐art methods. Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired. A denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high-dimension signals is more effective than that from the low-dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single-channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2-sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent. Experimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal-to-noise ratio, structural similarity, and high-frequency error norm. A simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state-of-the-art methods. Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired.PURPOSEAlthough recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired.A denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high-dimension signals is more effective than that from the low-dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single-channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2-sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent.METHODSA denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high-dimension signals is more effective than that from the low-dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single-channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2-sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent.Experimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal-to-noise ratio, structural similarity, and high-frequency error norm.RESULTSExperimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal-to-noise ratio, structural similarity, and high-frequency error norm.A simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state-of-the-art methods.CONCLUSIONA simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state-of-the-art methods.
Author	Cheng, Huitao Liu, Qiegen Liang, Dong Yang, Qingxin Wang, Shanshan Zhang, Minghui
Author_xml	– sequence: 1 givenname: Qiegen orcidid: 0000-0003-4717-2283 surname: Liu fullname: Liu, Qiegen organization: Nanchang University – sequence: 2 givenname: Qingxin surname: Yang fullname: Yang, Qingxin organization: Nanchang University – sequence: 3 givenname: Huitao surname: Cheng fullname: Cheng, Huitao organization: Chinese Academy of Sciences – sequence: 4 givenname: Shanshan orcidid: 0000-0002-0575-6523 surname: Wang fullname: Wang, Shanshan organization: Chinese Academy of Sciences – sequence: 5 givenname: Minghui surname: Zhang fullname: Zhang, Minghui organization: Nanchang University – sequence: 6 givenname: Dong orcidid: 0000-0003-0131-2519 surname: Liang fullname: Liang, Dong email: dong.liang@siat.ac.cn organization: Chinese Academy of Sciences
BackLink	https://www.ncbi.nlm.nih.gov/pubmed/31429993$$D View this record in MEDLINE/PubMed
BookMark	eNp9kU1Lw0AQhhep2A89-Ack4EUPafcrTfYoRa3QIojiMWw207ol2a27CdJ_7_ZDDwU9DfPyzDDvvH3UMdYAQpcEDwnGdFS7ekhTQckJ6pGE0pgmgndQD6ccx4wI3kV971cYYyFSfoa6jHAqhGA99D7Vy49qE7WmBOdlva6gjGq5NNBoFTnw1kijINJB02YZFGWNb1yrGm1N1PqtKNvGglG23DZrp63z5-h0ISsPF4c6QG8P96-TaTx7fnya3M1ixRJGYjkepzJTnElBIFFcplCCwBzwOC1UkpS8FJwt0gKYwGVBRCYzWohMBIfjTAAboJv93rWzny34Jq-1V1BV0oBtfU4ZJllGgu-AXh-hK9s6E67bUThJMGaBujpQbVFDmQc7tXSb_OdlAbjdA8pZ7x0sfhGC820ceYgj38UR2NERq3Qjt59rnNTVfxNfuoLN36vz-ct8P_ENaTubzA
CitedBy_id	crossref_primary_10_1109_TMI_2023_3288219 crossref_primary_10_1016_j_bspc_2025_108355 crossref_primary_10_1109_TNNLS_2023_3333538 crossref_primary_10_1109_TMI_2022_3180228 crossref_primary_10_1016_j_mri_2023_07_004 crossref_primary_10_1109_TMI_2023_3313421 crossref_primary_10_1109_TIM_2024_3417538 crossref_primary_10_1016_j_dsp_2025_105012 crossref_primary_10_1109_TNNLS_2021_3090303 crossref_primary_10_1016_j_compbiomed_2024_108668 crossref_primary_10_1109_JPROC_2022_3141367 crossref_primary_10_1016_j_jvcir_2020_102927 crossref_primary_10_1109_TMI_2022_3153849 crossref_primary_10_3390_electronics11040586 crossref_primary_10_1109_TCI_2022_3145187 crossref_primary_10_1016_j_inffus_2020_09_006 crossref_primary_10_1002_ima_70013 crossref_primary_10_1002_jbio_202300281 crossref_primary_10_3390_bioengineering10091012 crossref_primary_10_1109_TMI_2023_3234968 crossref_primary_10_3390_photonics9050343 crossref_primary_10_1016_j_dsp_2020_102856 crossref_primary_10_1002_nbm_4848 crossref_primary_10_1109_TMI_2024_3462988 crossref_primary_10_1002_mrm_29624 crossref_primary_10_1016_j_cmpb_2023_107590 crossref_primary_10_1016_j_compbiomed_2023_107610 crossref_primary_10_1093_jcde_qwae006 crossref_primary_10_3390_app15147717 crossref_primary_10_1109_JAS_2023_123543 crossref_primary_10_1016_j_mri_2023_02_004 crossref_primary_10_1016_j_mri_2020_02_002 crossref_primary_10_1016_j_sigpro_2021_108320 crossref_primary_10_1002_nbm_5005 crossref_primary_10_3390_electronics12234742 crossref_primary_10_1098_rsta_2024_0323 crossref_primary_10_1016_j_bspc_2025_108291 crossref_primary_10_1016_j_media_2023_102872 crossref_primary_10_1016_j_mri_2021_12_003 crossref_primary_10_1007_s00371_022_02747_0 crossref_primary_10_1109_TMI_2021_3096232 crossref_primary_10_3934_ipi_2020051 crossref_primary_10_1002_mrm_29085 crossref_primary_10_1016_j_neucom_2020_09_086 crossref_primary_10_1109_TMI_2022_3147426 crossref_primary_10_1016_j_bspc_2021_102579 crossref_primary_10_1109_TMM_2022_3177933 crossref_primary_10_1016_j_mri_2024_01_007 crossref_primary_10_1109_TMI_2021_3081677
Cites_doi	10.1109/CVPR.2018.00196 10.1137/17M1141771 10.1002/mrm.21391 10.1002/mrm.22736 10.1002/mrm.21236 10.1016/j.media.2013.09.007 10.1109/CVPRW.2017.152 10.1109/TIP.2017.2704443 10.1109/IEMBS.2004.1403345 10.1109/TMI.2013.2256464 10.1109/TMI.2010.2089519 10.1109/TMI.2010.2090538 10.1109/CVPR.2017.374 10.1109/34.1000236 10.1002/mrm.24267 10.1016/j.jvcir.2017.07.002 10.1145/1390156.1390294 10.1109/LNLA.2009.5278405 10.1109/TMI.2016.2550204 10.1137/090746379 10.1109/TIP.2003.819861 10.1109/TIP.2014.2329449 10.1109/TIP.2017.2662206 10.1002/mrm.22595 10.1109/ICIP.2017.8296620 10.1109/TIP.2009.2012908 10.1007/978-3-319-59050-9_51 10.1002/mrm.10718 10.1109/CVPRW.2017.151 10.1561/2400000003 10.1109/TMI.2018.2887072 10.1109/TBME.2015.2503756 10.1109/CVPR.2009.5206802 10.1109/ISBI.2016.7493320
ContentType	Journal Article
Copyright	2019 International Society for Magnetic Resonance in Medicine 2019 International Society for Magnetic Resonance in Medicine.
Copyright_xml	– notice: 2019 International Society for Magnetic Resonance in Medicine – notice: 2019 International Society for Magnetic Resonance in Medicine.
DBID	AAYXX CITATION CGR CUY CVF ECM EIF NPM 8FD FR3 K9. M7Z P64 7X8
DOI	10.1002/mrm.27921
DatabaseName	CrossRef Medline MEDLINE MEDLINE (Ovid) MEDLINE MEDLINE PubMed Technology Research Database Engineering Research Database ProQuest Health & Medical Complete (Alumni) Biochemistry Abstracts 1 Biotechnology and BioEngineering Abstracts MEDLINE - Academic
DatabaseTitle	CrossRef MEDLINE Medline Complete MEDLINE with Full Text PubMed MEDLINE (Ovid) Biochemistry Abstracts 1 ProQuest Health & Medical Complete (Alumni) Engineering Research Database Technology Research Database Biotechnology and BioEngineering Abstracts MEDLINE - Academic
DatabaseTitleList	Biochemistry Abstracts 1 MEDLINE MEDLINE - Academic
Database_xml	– sequence: 1 dbid: NPM name: PubMed url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 2 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database
DeliveryMethod	fulltext_linktorsrc
Discipline	Medicine Physics
EISSN	1522-2594
EndPage	336
ExternalDocumentID	31429993 10_1002_mrm_27921 MRM27921
Genre	article Research Support, Non-U.S. Gov't Journal Article
GrantInformation_xml	– fundername: Basic Research Program of Shenzhen funderid: JCYJ20150831154213680 – fundername: National Natural Science Foundation of China funderid: 61661031; 61871206 – fundername: Natural Science Foundation of Jiangxi Province funderid: 20181BAB202003
GroupedDBID	--- -DZ .3N .55 .GA .Y3 05W 0R~ 10A 1L6 1OB 1OC 1ZS 24P 31~ 33P 3O- 3SF 3WU 4.4 4ZD 50Y 50Z 51W 51X 52M 52N 52O 52P 52R 52S 52T 52U 52V 52W 52X 53G 5GY 5RE 5VS 66C 702 7PT 8-0 8-1 8-3 8-4 8-5 8UM 930 A01 A03 AAESR AAEVG AAHHS AAHQN AAIPD AAMNL AANHP AANLZ AAONW AASGY AAXRX AAYCA AAZKR ABCQN ABCUV ABDPE ABEML ABIJN ABJNI ABLJU ABPVW ABQWH ABXGK ACAHQ ACBWZ ACCFJ ACCZN ACFBH ACGFO ACGFS ACGOF ACIWK ACMXC ACPOU ACPRK ACRPL ACSCC ACXBN ACXQS ACYXJ ADBBV ADBTR ADEOM ADIZJ ADKYN ADMGS ADNMO ADOZA ADXAS ADZMN AEEZP AEGXH AEIGN AEIMD AENEX AEQDE AEUQT AEUYR AFBPY AFFNX AFFPM AFGKR AFPWT AFRAH AFWVQ AFZJQ AHBTC AHMBA AIACR AIAGR AITYG AIURR AIWBW AJBDE ALAGY ALMA_UNASSIGNED_HOLDINGS ALUQN ALVPJ AMBMR AMYDB ASPBG ATUGU AVWKF AZBYB AZFZN AZVAB BAFTC BDRZF BFHJK BHBCM BMXJE BROTX BRXPI BY8 C45 CS3 D-6 D-7 D-E D-F DCZOG DPXWK DR2 DRFUL DRMAN DRSTM DU5 EBD EBS EJD EMOBN F00 F01 F04 FEDTE FUBAC G-S G.N GNP GODZA H.X HBH HDBZQ HF~ HGLYW HHY HHZ HVGLF HZ~ I-F IX1 J0M JPC KBYEO KQQ LATKE LAW LC2 LC3 LEEKS LH4 LITHE LOXES LP6 LP7 LUTES LW6 LYRES M65 MEWTI MK4 MRFUL MRMAN MRSTM MSFUL MSMAN MSSTM MXFUL MXMAN MXSTM N04 N05 N9A NF~ NNB O66 O9- OIG OVD P2P P2W P2X P2Z P4B P4D PALCI PQQKQ Q.N Q11 QB0 QRW R.K RGB RIWAO RJQFR ROL RWI RX1 RYL SAMSI SUPJJ SV3 TEORI TUS TWZ UB1 V2E V8K W8V W99 WBKPD WHWMO WIB WIH WIJ WIK WIN WJL WOHZO WQJ WRC WUP WVDHM WXI WXSBR X7M XG1 XPP XV2 ZGI ZXP ZZTAW ~IA ~WT AAMMB AAYXX AEFGJ AEYWJ AGHNM AGQPQ AGXDD AGYGG AIDQK AIDYY AIQQE CITATION O8X CGR CUY CVF ECM EIF NPM 8FD FR3 K9. M7Z P64 7X8
ID	FETCH-LOGICAL-c3531-a667a8c43a91e5c4a7ede904e067bc55d4d943f7be390db198a82b989259689e3
IEDL.DBID	WIN
ISICitedReferencesCount	52
ISICitedReferencesURI	http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000482277500001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN	0740-3194 1522-2594
IngestDate	Wed Oct 01 15:01:36 EDT 2025 Sat Nov 29 14:27:12 EST 2025 Mon Jul 21 06:08:07 EDT 2025 Sat Nov 29 02:37:48 EST 2025 Tue Nov 18 22:12:39 EST 2025 Wed Jan 22 16:38:05 EST 2025
IsPeerReviewed	true
IsScholarly	true
Issue	1
Keywords	multichannel prior autoencoding priors magnetic resonance imaging image reconstruction proximal gradient descent
Language	English
License	2019 International Society for Magnetic Resonance in Medicine.
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c3531-a667a8c43a91e5c4a7ede904e067bc55d4d943f7be390db198a82b989259689e3
Notes	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 content type line 23
ORCID	0000-0002-0575-6523 0000-0003-0131-2519 0000-0003-4717-2283
PMID	31429993
PQID	2301055003
PQPubID	1016391
PageCount	15
ParticipantIDs	proquest_miscellaneous_2301881997 proquest_journals_2301055003 pubmed_primary_31429993 crossref_primary_10_1002_mrm_27921 crossref_citationtrail_10_1002_mrm_27921 wiley_primary_10_1002_mrm_27921_MRM27921
PublicationCentury	2000
PublicationDate	January 2020 2020-01-00 20200101
PublicationDateYYYYMMDD	2020-01-01
PublicationDate_xml	– month: 01 year: 2020 text: January 2020
PublicationDecade	2020
PublicationPlace	United States
PublicationPlace_xml	– name: United States – name: Hoboken
PublicationTitle	Magnetic resonance in medicine
PublicationTitleAlternate	Magn Reson Med
PublicationYear	2020
Publisher	Wiley Subscription Services, Inc
Publisher_xml	– name: Wiley Subscription Services, Inc
References	2010; 11 2013; 69 2017; 48 2017; 26 2011 2016; 32 2009 2008 2011; 30 2019; 38 2004 2007; 31 2007; 57 2014; 23 2007; 58 2014; 1 2004; 51 2013; 32 2017; 33 2002; 24 2004; 13 2014; 15 2018 2016; 63 2011; 65 2017 2016 2015 2014; 18 2010; 3 2018; 11 2009; 18 e_1_2_8_28_1 Alain G (e_1_2_8_39_1) 2014; 15 e_1_2_8_24_1 e_1_2_8_47_1 e_1_2_8_26_1 e_1_2_8_49_1 Sun J (e_1_2_8_25_1) 2016 Mao XJ (e_1_2_8_42_1) 2016 e_1_2_8_3_1 Ma S (e_1_2_8_7_1) 2008 e_1_2_8_5_1 e_1_2_8_9_1 e_1_2_8_20_1 e_1_2_8_43_1 e_1_2_8_22_1 e_1_2_8_45_1 e_1_2_8_41_1 e_1_2_8_19_1 e_1_2_8_13_1 e_1_2_8_15_1 He K (e_1_2_8_17_1) 2015 e_1_2_8_32_1 e_1_2_8_11_1 e_1_2_8_34_1 e_1_2_8_30_1 Vincent P (e_1_2_8_36_1) 2010; 11 Qu P (e_1_2_8_4_1) 2007; 31 e_1_2_8_29_1 Glorot X (e_1_2_8_37_1) 2011 e_1_2_8_46_1 e_1_2_8_48_1 e_1_2_8_2_1 e_1_2_8_6_1 e_1_2_8_8_1 e_1_2_8_23_1 e_1_2_8_44_1 e_1_2_8_40_1 e_1_2_8_18_1 e_1_2_8_14_1 Jonas A (e_1_2_8_27_1) 2017; 33 e_1_2_8_35_1 e_1_2_8_16_1 Hammernik K (e_1_2_8_21_1) 2016 e_1_2_8_10_1 e_1_2_8_31_1 Geras KJ (e_1_2_8_38_1) 2015 e_1_2_8_12_1 e_1_2_8_33_1 e_1_2_8_50_1
References_xml	– volume: 63 start-page: 1850 year: 2016 end-page: 1861 article-title: Fast multiclass dictionaries learning with geometrical directions in MRI reconstruction publication-title: IEEE Trans Biomed Eng – volume: 30 start-page: 1090 year: 2011 end-page: 1099 article-title: Compressed sensing with wavelet domain dependencies for coronary MRI: Aaretrospective study” publication-title: IEEE Trans Med Imaging – volume: 69 start-page: 571 year: 2013 end-page: 582 article-title: Coil compression for accelerated imaging with Cartesian sampling publication-title: Magn Reson Med – volume: 23 start-page: 3618 year: 2014 end-page: 3632 article-title: Compressive sensing via nonlocal low‐rank regularization publication-title: IEEE Trans Image Process – volume: 51 start-page: 559 year: 2004 end-page: 567 article-title: Parallel imaging reconstruction using automatic regularization publication-title: Magn Reson Med – volume: 57 start-page: 1086 year: 2007 end-page: 1098 article-title: Undersampled radial MRI with multiple coils. Iterative image reconstruction using a total variation constraint publication-title: Magn Reson Med – volume: 13 start-page: 600 year: 2004 end-page: 612 article-title: Image quality assessment: from error visibility to structural similarity publication-title: IEEE Trans Image Process – start-page: 136 year: 2017 end-page: 144 – start-page: 315 year: 2011 end-page: 323 – volume: 3 start-page: 253 year: 2010 end-page: 276 article-title: Bregmanized nonlocal regularization for deconvolution and sparse reconstruction publication-title: SIAM J Imaging Sci – volume: 26 start-page: 5094 year: 2017 end-page: 5106 article-title: Sparsity‐based color image super resolution via exploiting cross channel constraints publication-title: IEEE Trans Image Process – volume: 18 start-page: 969 year: 2009 end-page: 981 article-title: Softcuts: a soft edge smoothness prior for color image super‐resolution publication-title: IEEE Trans Image Process – volume: 30 start-page: 1028 year: 2011 end-page: 1041 article-title: MR image reconstruction from highly undersampled k‐space data by dictionary learning publication-title: IEEE Trans Med Imaging – start-page: 2802 year: 2016 end-page: 2810 – start-page: 145 year: 2017 end-page: 153 – volume: 15 start-page: 3743 year: 2014 end-page: 3773 article-title: What regularized auto‐encoders learn from the data‐generating distribution publication-title: JMLR – volume: 31 start-page: 44 year: 2007 end-page: 50 article-title: An improved iterative SENSE reconstruction method publication-title: Magn Reson Eng – volume: 65 start-page: 1384 year: 2011 end-page: 1392 article-title: Sensitivity encoding reconstruction with nonlocal total variation regularization publication-title: Magn Reson Med – volume: 33 start-page: 1 year: 2017 end-page: 24 article-title: Solving ill‐posed inverse problem using iterative deep neural networks publication-title: Inverse Prob – volume: 11 start-page: 3371 year: 2010 end-page: 3408 article-title: Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion publication-title: JMLR – volume: 58 start-page: 1182 year: 2007 end-page: 1195 article-title: Sparse MRI: the application of compressed sensing for rapid MR imaging publication-title: Magn Reson Med – volume: 1 start-page: 127 year: 2014 end-page: 239 article-title: Proximal algorithms publication-title: Found. Trends Optim – start-page: 1828 year: 2018 end-page: 1837 – start-page: 1088 year: 2016 – start-page: 1 year: 2008 end-page: 8 – volume: 26 start-page: 3142 year: 2017 end-page: 3155 article-title: Beyond a Gaussian denoiser: residual learning of deep CNN for image denoising publication-title: IEEE Trans Image Process – start-page: 770 year: 2015 end-page: 778 – start-page: 10 year: 2016 end-page: 18 – start-page: 1550 year: 2009 end-page: 1557 – year: 2008 – volume: 32 start-page: 1290 year: 2013 end-page: 1301 article-title: Highly undersampled magnetic resonance imaging reconstruction using two‐level Bregman method with dictionary updating publication-title: IEEE Trans Med Imaging – volume: 11 start-page: 991 year: 2018 end-page: 1048 article-title: Deep convolutional framelets: a general deep learning framework for inverse problems publication-title: SIAM J Imaging Sci – start-page: 647 year: 2017 end-page: 658 – volume: 48 start-page: 268 year: 2017 end-page: 280 article-title: A two‐stage convolutional sparse prior model for image restoration publication-title: J Vis Commun Image R – start-page: 4467 year: 2017 end-page: 4477 – volume: 18 start-page: 843 year: 2014 end-page: 856 article-title: Magnetic resonance image reconstruction from undersampled measurements using a patch‐based nonlocal operator publication-title: Med Image Anal – start-page: 46 year: 2009 end-page: 55 – start-page: 1940 year: 2017 end-page: 1944 – volume: 32 start-page: 2119 year: 2016 end-page: 2129 article-title: Accelerated high‐dimensional MR imaging with sparse sampling using low‐rank tensors publication-title: IEEE Trans Med Imaging – year: 2017 – volume: 38 start-page: 99 year: 2019 article-title: MR image reconstruction using deep density priors publication-title: IEEE Trans Med Imaging – start-page: 514 year: 2016 end-page: 517 – year: 2015 – volume: 65 start-page: 480 year: 2011 end-page: 491 article-title: Second order total generalized variation (TGV) for MRI publication-title: Magn Reson Med – volume: 24 start-page: 603 year: 2002 end-page: 619 article-title: Mean shift: a robust approach toward feature space analysis publication-title: IEEE Trans. Pattern Anal Mach Intell – start-page: 1056 year: 2004 end-page: 1059 – ident: e_1_2_8_28_1 doi: 10.1109/CVPR.2018.00196 – ident: e_1_2_8_24_1 doi: 10.1137/17M1141771 – ident: e_1_2_8_2_1 doi: 10.1002/mrm.21391 – ident: e_1_2_8_10_1 doi: 10.1002/mrm.22736 – ident: e_1_2_8_6_1 doi: 10.1002/mrm.21236 – start-page: 10 volume-title: Proceedings of the Thirtieth Conference on Neural Information Processing Systems (NIPS) year: 2016 ident: e_1_2_8_25_1 – ident: e_1_2_8_32_1 – ident: e_1_2_8_12_1 doi: 10.1016/j.media.2013.09.007 – ident: e_1_2_8_18_1 doi: 10.1109/CVPRW.2017.152 – start-page: 315 volume-title: Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, PMLR year: 2011 ident: e_1_2_8_37_1 – ident: e_1_2_8_45_1 doi: 10.1109/TIP.2017.2704443 – ident: e_1_2_8_5_1 doi: 10.1109/IEMBS.2004.1403345 – volume: 15 start-page: 3743 year: 2014 ident: e_1_2_8_39_1 article-title: What regularized auto‐encoders learn from the data‐generating distribution publication-title: JMLR – ident: e_1_2_8_13_1 doi: 10.1109/TMI.2013.2256464 – ident: e_1_2_8_8_1 doi: 10.1109/TMI.2010.2089519 – ident: e_1_2_8_11_1 doi: 10.1109/TMI.2010.2090538 – ident: e_1_2_8_40_1 doi: 10.1109/CVPR.2017.374 – ident: e_1_2_8_41_1 doi: 10.1109/34.1000236 – start-page: 1088 volume-title: Proceedings of the 24th Annual Meeting & Exhibition of the ISMRM year: 2016 ident: e_1_2_8_21_1 – ident: e_1_2_8_43_1 doi: 10.1002/mrm.24267 – ident: e_1_2_8_14_1 doi: 10.1016/j.jvcir.2017.07.002 – ident: e_1_2_8_35_1 doi: 10.1145/1390156.1390294 – ident: e_1_2_8_33_1 doi: 10.1109/LNLA.2009.5278405 – ident: e_1_2_8_16_1 doi: 10.1109/TMI.2016.2550204 – ident: e_1_2_8_23_1 – ident: e_1_2_8_29_1 – ident: e_1_2_8_34_1 doi: 10.1137/090746379 – volume: 11 start-page: 3371 year: 2010 ident: e_1_2_8_36_1 article-title: Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion publication-title: JMLR – ident: e_1_2_8_50_1 doi: 10.1109/TIP.2003.819861 – ident: e_1_2_8_26_1 – volume-title: Proceedings of the 3rd International Conference on Learning Representations (ICLR) year: 2015 ident: e_1_2_8_38_1 – start-page: 2802 volume-title: Proceedings of the Thirtieth Conference on Neural Information Processing Systems (NIPS) year: 2016 ident: e_1_2_8_42_1 – ident: e_1_2_8_49_1 doi: 10.1109/TIP.2014.2329449 – ident: e_1_2_8_31_1 doi: 10.1109/TIP.2017.2662206 – ident: e_1_2_8_9_1 doi: 10.1002/mrm.22595 – ident: e_1_2_8_15_1 doi: 10.1109/ICIP.2017.8296620 – volume: 31 start-page: 44 year: 2007 ident: e_1_2_8_4_1 article-title: An improved iterative SENSE reconstruction method publication-title: Magn Reson Eng – ident: e_1_2_8_46_1 doi: 10.1109/TIP.2009.2012908 – ident: e_1_2_8_22_1 doi: 10.1007/978-3-319-59050-9_51 – ident: e_1_2_8_3_1 doi: 10.1002/mrm.10718 – ident: e_1_2_8_19_1 doi: 10.1109/CVPRW.2017.151 – volume: 33 start-page: 1 year: 2017 ident: e_1_2_8_27_1 article-title: Solving ill‐posed inverse problem using iterative deep neural networks publication-title: Inverse Prob – ident: e_1_2_8_47_1 doi: 10.1561/2400000003 – ident: e_1_2_8_30_1 doi: 10.1109/TMI.2018.2887072 – ident: e_1_2_8_48_1 doi: 10.1109/TBME.2015.2503756 – ident: e_1_2_8_44_1 doi: 10.1109/CVPR.2009.5206802 – ident: e_1_2_8_20_1 doi: 10.1109/ISBI.2016.7493320 – start-page: 1 volume-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) year: 2008 ident: e_1_2_8_7_1 – start-page: 770 volume-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) year: 2015 ident: e_1_2_8_17_1
SSID	ssj0009974
Score	2.5567253
Snippet	Purpose Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage... Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into... PurposeAlthough recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage...
SourceID	proquest pubmed crossref wiley
SourceType	Aggregation Database Index Database Enrichment Source Publisher
StartPage	322
SubjectTerms	Acceleration Algorithms autoencoding priors Brain - diagnostic imaging Computer Systems Humans Image processing Image Processing, Computer-Assisted - methods Image reconstruction Image restoration Machine learning Magnetic Resonance Imaging multichannel prior Neural Networks, Computer Noise generation Noise reduction proximal gradient descent Sampling Signal-To-Noise Ratio Software Trajectories
Title	Highly undersampled magnetic resonance imaging reconstruction using autoencoding priors
URI	https://onlinelibrary.wiley.com/doi/abs/10.1002%2Fmrm.27921 https://www.ncbi.nlm.nih.gov/pubmed/31429993 https://www.proquest.com/docview/2301055003 https://www.proquest.com/docview/2301881997
Volume	83
WOSCitedRecordID	wos000482277500001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText	1
inHoldings	1
isFullTextHit
isPrint
journalDatabaseRights	– providerCode: PRVWIB databaseName: Wiley Online Library - Journals customDbUrl: eissn: 1522-2594 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0009974 issn: 0740-3194 databaseCode: DRFUL dateStart: 19990101 isFulltext: true titleUrlDefault: https://onlinelibrary.wiley.com providerName: Wiley-Blackwell – providerCode: PRVWIB databaseName: Wiley Online Library Free Content customDbUrl: eissn: 1522-2594 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0009974 issn: 0740-3194 databaseCode: WIN dateStart: 19990101 isFulltext: true titleUrlDefault: https://onlinelibrary.wiley.com providerName: Wiley-Blackwell
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV3da9swED-y7oO97KPdR9Y0aKMPefFqW7Il0afSLXSwhBEWmjcjS2oJNHawk8H---lkxyVsg8HeDD4ho9Ppd2fd_Q7gVGkbpTbOA5EoFTCR60AYaQIHLpSrGyYUb5pN8OlULBbyWw_Od7UwDT9E98MNLcOf12jgKq_P7klDV9XqI7LfYegTMW-U11-m94S7smFg5gzPGcl2rEJhfNaN3Mei3xzMfX_VA874-X996gt41vqZ5KLZGC-hZ4tDeDJpb9IP4bFP_dT1EVxjqsfdT7L1ZS4K2YINWanbAssbiYvGS-TksGS58g2NiA-hO9pZgonzt0RtNyVSYiISknW1LKv6FczHn79fXgVtu4VAU2eJgUpTroRmVMnIJpopbo2VIbMO0HKdJIYZyegNzy2VockjKZSIcymki6BSIS19DQdFWdi3QCRNqU4d8GHxCYucT5QbF8ml1IZcRkb1YbRb-Ey3XOTYEuMua1iU48wtWeaXrA8fOtF1Q8DxJ6HBTntZa4N15oIr7P7pjq0-vO9eO-vBKxFV2HLbyAiByTZ9eNNovZuFRojV0o0eeeX-ffpsMpv4h3f_LnoMT2MM3f3fnAEcOLXZE3ikf2yWdTWEB3whhvDw02w8_zr0O_sXKpj4iw
linkProvider	Wiley-Blackwell
linkToHtml	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV3dS8MwED-G3y9-f0ynRvHBl-raZG0Cvog4Jm5DRNG3kiZRBq6VzQn-9-bSrjJUEHwr9EJLLpe7X3L3O4AjqYwfmiDxeENKj_FEeVwL7VnnQiP5xLiM8mYTUbfLHx_FTQXOxrUwOT9EeeCGluH2azRwPJA-_WIN7Q_6J0h_Z7HPNAuZwP4FD1fdL8pdkXMwRwx3GsHGvEL14LQcOumNvoWYkxGrcznNpf_97DIsFqEmOc_XxgpUTLoKc53iMn0VZl32pxquwQNme7x8kJGrdJFIGKxJXz6nWOFILCDPkJbDkF7f9TQiDkWXzLMEc-efiRy9ZciKic6QvA562WC4DvfNy7uLlld0XPAUtcboyTCMJFeMSuGbhmIyMtqIOjPWpyWq0dBMC0afosRQUdeJL7jkQSK4sCAq5MLQDZhKs9RsARE0pCq0vg_rT5hvw6JEWzAXUlOPhK9lFY7HMx-rgo4cu2K8xDmRchDbKYvdlFXhsBR9zTk4fhKqjdUXF2Y4jC2-wgagdueqwkH52hoQ3orI1GSjXIZzzLepwmau9vIr1Ed3LezoY6fd3z8fd2477mH776L7MN-667Tj9lX3egcWAkTy7nCnBlNWhWYXZtT7W2842HML-xMpUvos
linkToPdf	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV3dS8MwED9kfuCL3x_zM4oPvlTXJm0T8EXUoeiGiKJvJU0yGbh2bE7wvzeXdhVRQfCt0AspuVx-d83d7wAOpDJ-ZILU46GUHuOp8rgW2rPgQmPZYVzGRbOJuN3mT0_idgJOxrUwBT9E9cMNLcOd12jgpq87x5-sob1B7wjp72zsM8lCEbIaTJ7fNR9uPkl3RcHCHDM8awQbMws1guNq8Fc8-uZkfvVZHeg05__3uQswVzqb5LTYHYswYbIlmGmV1-lLMO3yP9VwGR4x3-PlnYxcrYtEymBNevI5wxpHYkPyHIk5DOn2XFcj4uLoinuWYPb8M5Gj1xx5MREOSX_QzQfDFXhoXtyfXXplzwVPUWuOnoyiWHLFqBS-CRWTsdFGNJixqJaqMNRMC0Y7cWqoaOjUF1zyIBVc2DAq4sLQVahleWbWgQgaURVZ9MMKFOZbxyjVNpyLqGnEwteyDofjlU9USUiOfTFekoJKOUjskiVuyeqwX4n2CxaOn4S2xupLSkMcJjbCwhag9uyqw1712poQ3ovIzOSjQoZzzLipw1qh9moW6iNgCzv60Gn39-mT1l3LPWz8XXQXZm7Pm8nNVft6E2YDDOXd350tqFkNmm2YUm-v3eFgp9zZH549-sw
openUrl	ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Highly+undersampled+magnetic+resonance+imaging+reconstruction+using+autoencoding+priors&rft.jtitle=Magnetic+resonance+in+medicine&rft.au=Liu%2C+Qiegen&rft.au=Yang%2C+Qingxin&rft.au=Cheng%2C+Huitao&rft.au=Wang%2C+Shanshan&rft.date=2020-01-01&rft.eissn=1522-2594&rft.volume=83&rft.issue=1&rft.spage=322&rft_id=info:doi/10.1002%2Fmrm.27921&rft_id=info%3Apmid%2F31429993&rft.externalDocID=31429993
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0740-3194&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0740-3194&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0740-3194&client=summon