Small training dataset convolutional neural networks for application-specific super-resolution microscopy

Machine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to trai...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of biomedical optics Ročník 28; číslo 3; s. 036501
Hlavní autoři: Mannam, Varun, Howard, Scott
Médium: Journal Article
Jazyk:angličtina
Vydáno: United States Society of Photo-Optical Instrumentation Engineers 01.03.2023
SPIE
S P I E - International Society for
Témata:
ISSN:1083-3668, 1560-2281, 1560-2281
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Abstract Machine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to train the neural network. We demonstrate how adding a "dense encoder-decoder" (DenseED) block can be used to effectively train a neural network that produces super-resolution (SR) images from conventional microscopy diffraction-limited (DL) images trained using a small dataset [15 fields of view (FOVs)]. The ML helps to retrieve SR information from a DL image when trained with a massive training dataset. The aim of this work is to demonstrate a neural network that estimates SR images from DL images using modifications that enable training with a small dataset. We employ "DenseED" blocks in existing SR ML network architectures. DenseED blocks use a dense layer that concatenates features from the previous convolutional layer to the next convolutional layer. DenseED blocks in fully convolutional networks (FCNs) estimate the SR images when trained with a small training dataset (15 FOVs) of human cells from the Widefield2SIM dataset and in fluorescent-labeled fixed bovine pulmonary artery endothelial cells samples. Conventional ML models without DenseED blocks trained on small datasets fail to accurately estimate SR images while models including the DenseED blocks can. The average peak SNR (PSNR) and resolution improvements achieved by networks containing DenseED blocks are and , respectively. We evaluated various configurations of target image generation methods (e.g., experimentally captured a target and computationally generated target) that are used to train FCNs with and without DenseED blocks and showed that including DenseED blocks in simple FCNs outperforms compared to simple FCNs without DenseED blocks. DenseED blocks in neural networks show accurate extraction of SR images even if the ML model is trained with a small training dataset of 15 FOVs. This approach shows that microscopy applications can use DenseED blocks to train on smaller datasets that are application-specific imaging platforms and there is promise for applying this to other imaging modalities, such as MRI/x-ray, etc.
AbstractList SignificanceMachine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to train the neural network. We demonstrate how adding a “dense encoder-decoder” (DenseED) block can be used to effectively train a neural network that produces super-resolution (SR) images from conventional microscopy diffraction-limited (DL) images trained using a small dataset [15 fields of view (FOVs)].AimThe ML helps to retrieve SR information from a DL image when trained with a massive training dataset. The aim of this work is to demonstrate a neural network that estimates SR images from DL images using modifications that enable training with a small dataset.ApproachWe employ “DenseED” blocks in existing SR ML network architectures. DenseED blocks use a dense layer that concatenates features from the previous convolutional layer to the next convolutional layer. DenseED blocks in fully convolutional networks (FCNs) estimate the SR images when trained with a small training dataset (15 FOVs) of human cells from the Widefield2SIM dataset and in fluorescent-labeled fixed bovine pulmonary artery endothelial cells samples.ResultsConventional ML models without DenseED blocks trained on small datasets fail to accurately estimate SR images while models including the DenseED blocks can. The average peak SNR (PSNR) and resolution improvements achieved by networks containing DenseED blocks are ≈3.2 dB and 2 × , respectively. We evaluated various configurations of target image generation methods (e.g., experimentally captured a target and computationally generated target) that are used to train FCNs with and without DenseED blocks and showed that including DenseED blocks in simple FCNs outperforms compared to simple FCNs without DenseED blocks.ConclusionsDenseED blocks in neural networks show accurate extraction of SR images even if the ML model is trained with a small training dataset of 15 FOVs. This approach shows that microscopy applications can use DenseED blocks to train on smaller datasets that are application-specific imaging platforms and there is promise for applying this to other imaging modalities, such as MRI/x-ray, etc.
Machine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to train the neural network. We demonstrate how adding a "dense encoder-decoder" (DenseED) block can be used to effectively train a neural network that produces super-resolution (SR) images from conventional microscopy diffraction-limited (DL) images trained using a small dataset [15 fields of view (FOVs)]. The ML helps to retrieve SR information from a DL image when trained with a massive training dataset. The aim of this work is to demonstrate a neural network that estimates SR images from DL images using modifications that enable training with a small dataset. We employ "DenseED" blocks in existing SR ML network architectures. DenseED blocks use a dense layer that concatenates features from the previous convolutional layer to the next convolutional layer. DenseED blocks in fully convolutional networks (FCNs) estimate the SR images when trained with a small training dataset (15 FOVs) of human cells from the Widefield2SIM dataset and in fluorescent-labeled fixed bovine pulmonary artery endothelial cells samples. Conventional ML models without DenseED blocks trained on small datasets fail to accurately estimate SR images while models including the DenseED blocks can. The average peak SNR (PSNR) and resolution improvements achieved by networks containing DenseED blocks are and , respectively. We evaluated various configurations of target image generation methods (e.g., experimentally captured a target and computationally generated target) that are used to train FCNs with and without DenseED blocks and showed that including DenseED blocks in simple FCNs outperforms compared to simple FCNs without DenseED blocks. DenseED blocks in neural networks show accurate extraction of SR images even if the ML model is trained with a small training dataset of 15 FOVs. This approach shows that microscopy applications can use DenseED blocks to train on smaller datasets that are application-specific imaging platforms and there is promise for applying this to other imaging modalities, such as MRI/x-ray, etc.
Machine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to train the neural network. We demonstrate how adding a "dense encoder-decoder" (DenseED) block can be used to effectively train a neural network that produces super-resolution (SR) images from conventional microscopy diffraction-limited (DL) images trained using a small dataset [15 fields of view (FOVs)].SignificanceMachine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise ratio (SNR)], and data interpretation. The bottleneck in developing effective ML systems is often the need to acquire large datasets to train the neural network. We demonstrate how adding a "dense encoder-decoder" (DenseED) block can be used to effectively train a neural network that produces super-resolution (SR) images from conventional microscopy diffraction-limited (DL) images trained using a small dataset [15 fields of view (FOVs)].The ML helps to retrieve SR information from a DL image when trained with a massive training dataset. The aim of this work is to demonstrate a neural network that estimates SR images from DL images using modifications that enable training with a small dataset.AimThe ML helps to retrieve SR information from a DL image when trained with a massive training dataset. The aim of this work is to demonstrate a neural network that estimates SR images from DL images using modifications that enable training with a small dataset.We employ "DenseED" blocks in existing SR ML network architectures. DenseED blocks use a dense layer that concatenates features from the previous convolutional layer to the next convolutional layer. DenseED blocks in fully convolutional networks (FCNs) estimate the SR images when trained with a small training dataset (15 FOVs) of human cells from the Widefield2SIM dataset and in fluorescent-labeled fixed bovine pulmonary artery endothelial cells samples.ApproachWe employ "DenseED" blocks in existing SR ML network architectures. DenseED blocks use a dense layer that concatenates features from the previous convolutional layer to the next convolutional layer. DenseED blocks in fully convolutional networks (FCNs) estimate the SR images when trained with a small training dataset (15 FOVs) of human cells from the Widefield2SIM dataset and in fluorescent-labeled fixed bovine pulmonary artery endothelial cells samples.Conventional ML models without DenseED blocks trained on small datasets fail to accurately estimate SR images while models including the DenseED blocks can. The average peak SNR (PSNR) and resolution improvements achieved by networks containing DenseED blocks are ≈ 3.2    dB and 2 × , respectively. We evaluated various configurations of target image generation methods (e.g., experimentally captured a target and computationally generated target) that are used to train FCNs with and without DenseED blocks and showed that including DenseED blocks in simple FCNs outperforms compared to simple FCNs without DenseED blocks.ResultsConventional ML models without DenseED blocks trained on small datasets fail to accurately estimate SR images while models including the DenseED blocks can. The average peak SNR (PSNR) and resolution improvements achieved by networks containing DenseED blocks are ≈ 3.2    dB and 2 × , respectively. We evaluated various configurations of target image generation methods (e.g., experimentally captured a target and computationally generated target) that are used to train FCNs with and without DenseED blocks and showed that including DenseED blocks in simple FCNs outperforms compared to simple FCNs without DenseED blocks.DenseED blocks in neural networks show accurate extraction of SR images even if the ML model is trained with a small training dataset of 15 FOVs. This approach shows that microscopy applications can use DenseED blocks to train on smaller datasets that are application-specific imaging platforms and there is promise for applying this to other imaging modalities, such as MRI/x-ray, etc.ConclusionsDenseED blocks in neural networks show accurate extraction of SR images even if the ML model is trained with a small training dataset of 15 FOVs. This approach shows that microscopy applications can use DenseED blocks to train on smaller datasets that are application-specific imaging platforms and there is promise for applying this to other imaging modalities, such as MRI/x-ray, etc.
Audience Academic
Author Howard, Scott
Mannam, Varun
Author_xml – sequence: 1
  givenname: Varun
  orcidid: 0000-0002-1866-6092
  surname: Mannam
  fullname: Mannam, Varun
  email: vmannam@nd.edu
  organization: University of Notre Dame, Department of Electrical Engineering, Notre Dame, Indiana, United States
– sequence: 2
  givenname: Scott
  surname: Howard
  fullname: Howard, Scott
  email: vmannam@nd.edu
  organization: University of Notre Dame, Department of Electrical Engineering, Notre Dame, Indiana, United States
BackLink https://www.ncbi.nlm.nih.gov/pubmed/36925620$$D View this record in MEDLINE/PubMed
BookMark eNp9kl1v1SAYxxsz4170A3hjmnjjTSsPtBSuzLa4qVmyC_WaUApHJoUK7cy-vXTnuLlFDQkQ-P2f98Nizwevi-IloBoAurdQfzq5rDGrSY0IbRE8KQ6gpajCmMFeviNGKkIp2y8OU7pCCDHK6bNin1COW4rRQWE_j9K5co7Seus35SBnmfRcquCvg1tmG7x0pddLvD3mnyF-T6UJsZTT5KySK1GlSStrrCrTMulYRZ122nK0KoakwnTzvHhqpEv6xe48Kr6evf9y-qG6uDz_eHp8UamWNHOlOo4lZyBZ2_QD7mVHqTR5Y6TTlKKeNZhrw_uGNUZTjGEYEJJEAuMGE0OOindbu9PSj3pQ2ufknJiiHWW8EUFa8fDH229iE64FIAQEOMkW3uwsxPBj0WkWo01KOye9DksSuGOs4xQwZPT1I_QqLDGXLFOMYkI4Zu09tZFOC-tNyI7ValQcs6bFHHC3uq3_QuU16FzF3Hlj8_sDwas_M71L8Xd3MwBbYO1BitrcIYDEOkECRJ6gHKogYjtBWdM90ig733Z5nRH3X-Uu_jRZfV-Hfwt-ARPa2ZU
CitedBy_id crossref_primary_10_1007_s12652_024_04751_9
Cites_doi 10.1117/12.2578519
10.1109/CVPR46437.2021.00988
10.1109/TPAMI.2020.2982166
10.1109/TIP.2017.2662206
10.1142/S1793545822300166
10.1364/OE.26.033166
10.1073/pnas.1107547108
10.1098/rsta.2021.0110
10.1038/ncomms12471
10.1109/LSP.2021.3098230
10.1007/978-3-030-66415-2_31
10.1093/bioinformatics/bty241
10.1364/BOE.9.004077
10.1038/s41598-018-19426-7
10.1007/978-3-030-15712-8_59
10.1117/12.2578574
10.1364/BOE.10.001044
10.1073/pnas.0406877102
10.1109/BigData.2017.8258006
10.3389/fphy.2020.00118
10.1017/S1431927622000782
10.1109/CVPR.2018.00333
10.1109/ICCV48922.2021.00663
10.1109/ICCV.2001.937655
10.26599/BDMA.2018.9020001
10.1103/PhysRevLett.99.228105
10.1007/s10489-020-01902-1
10.1038/s41598-019-42294-8
10.1002/mp.12344
10.1002/mp.13717
10.1007/BF02956173
10.1007/978-3-319-24574-4_28
10.1109/CVPRW.2017.150
10.1038/ncomms13752
10.1038/s41592-018-0239-0
10.1016/j.jcp.2018.04.018
10.1007/s11554-021-01127-6
10.1007/s11042-019-7397-7
10.1364/JOSAA.12.000058
10.1186/s43074-022-00077-x
10.1083/jcb.200903097
10.1117/12.2607515
10.1038/s41467-022-30949-6
10.1109/TCI.2020.3031070
10.1046/j.1365-2818.2000.00710.x
10.1117/12.2608910
10.1007/s10278-017-0033-z
10.1007/s00330-021-08320-y
10.1038/s41587-021-01092-2
10.1364/OPTICA.5.000458
10.1038/nmeth929
10.1364/OPTICA.426870
10.1364/OL.19.000780
10.1364/OPTICA.448287
10.1109/TIP.2003.819861
10.1038/nmeth817
10.1016/j.bpj.2009.08.002
10.1088/1361-6463/ab0261
10.1007/s11760-019-01569-3
10.1038/nbt.4106
10.1126/science.1127344
10.1109/TBME.2018.2814538
10.1038/s41592-018-0216-7
10.1088/2515-7647/abac1a
ContentType Journal Article
Copyright The Authors. Published by SPIE under a Creative Commons Attribution 4.0 International License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.
2023 The Authors.
COPYRIGHT 2023 SPIE
2023. This work is licensed under https://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
2023 The Authors 2023 The Authors
Copyright_xml – notice: The Authors. Published by SPIE under a Creative Commons Attribution 4.0 International License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.
– notice: 2023 The Authors.
– notice: COPYRIGHT 2023 SPIE
– notice: 2023. This work is licensed under https://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
– notice: 2023 The Authors 2023 The Authors
DBID AAYXX
CITATION
CGR
CUY
CVF
ECM
EIF
NPM
7QF
7QO
7QQ
7SC
7SE
7SP
7SR
7TA
7TB
7U5
8BQ
8FD
8FE
8FH
ABUWG
AEUYN
AFKRA
AZQEC
BBNVY
BENPR
BHPHI
CCPQU
DWQXO
F28
FR3
GNUQQ
H8D
H8G
HCIFZ
JG9
JQ2
KR7
L7M
LK8
L~C
L~D
M7P
P64
PHGZM
PHGZT
PIMPY
PKEHL
PQEST
PQGLB
PQQKQ
PQUKI
7X8
5PM
DOI 10.1117/1.JBO.28.3.036501
DatabaseName CrossRef
Medline
MEDLINE
MEDLINE (Ovid)
MEDLINE
MEDLINE
PubMed
Aluminium Industry Abstracts
Biotechnology Research Abstracts
Ceramic Abstracts
Computer and Information Systems Abstracts
Corrosion Abstracts
Electronics & Communications Abstracts
Engineered Materials Abstracts
Materials Business File
Mechanical & Transportation Engineering Abstracts
Solid State and Superconductivity Abstracts
METADEX
Technology Research Database
ProQuest SciTech Collection
ProQuest Natural Science Collection
ProQuest Central (Alumni)
ProQuest One Sustainability (subscription)
ProQuest Central UK/Ireland
ProQuest Central Essentials
Biological Science Collection (subscription)
ProQuest Central
Natural Science Collection
ProQuest One Community College
ProQuest Central
ANTE: Abstracts in New Technology & Engineering
Engineering Research Database
ProQuest Central Student
Aerospace Database
Copper Technical Reference Library
SciTech Premium Collection
Materials Research Database
ProQuest Computer Science Collection
Civil Engineering Abstracts
Advanced Technologies Database with Aerospace
ProQuest Biological Science Collection
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
Biological Science Database
Biotechnology and BioEngineering Abstracts
ProQuest Central Premium
ProQuest One Academic (New)
Publicly Available Content Database (subscription)
ProQuest One Academic Middle East (New)
ProQuest One Academic Eastern Edition (DO NOT USE)
ProQuest One Applied & Life Sciences
ProQuest One Academic (retired)
ProQuest One Academic UKI Edition
MEDLINE - Academic
PubMed Central (Full Participant titles)
DatabaseTitle CrossRef
MEDLINE
Medline Complete
MEDLINE with Full Text
PubMed
MEDLINE (Ovid)
Publicly Available Content Database
Materials Research Database
ProQuest Central Student
Technology Research Database
Computer and Information Systems Abstracts – Academic
ProQuest One Academic Middle East (New)
Mechanical & Transportation Engineering Abstracts
ProQuest Central Essentials
ProQuest Computer Science Collection
Computer and Information Systems Abstracts
ProQuest Central (Alumni Edition)
SciTech Premium Collection
ProQuest One Community College
ProQuest Natural Science Collection
Materials Business File
ProQuest Central
ProQuest One Applied & Life Sciences
Aerospace Database
Copper Technical Reference Library
ProQuest One Sustainability
Engineered Materials Abstracts
Biotechnology Research Abstracts
Natural Science Collection
ProQuest Central Korea
Biological Science Collection
ProQuest Central (New)
Advanced Technologies Database with Aerospace
ANTE: Abstracts in New Technology & Engineering
Civil Engineering Abstracts
Aluminium Industry Abstracts
ProQuest Biological Science Collection
ProQuest One Academic Eastern Edition
Electronics & Communications Abstracts
Ceramic Abstracts
Biological Science Database
ProQuest SciTech Collection
METADEX
Biotechnology and BioEngineering Abstracts
Computer and Information Systems Abstracts Professional
ProQuest One Academic UKI Edition
Solid State and Superconductivity Abstracts
Engineering Research Database
ProQuest One Academic
Corrosion Abstracts
ProQuest One Academic (New)
MEDLINE - Academic
DatabaseTitleList Publicly Available Content Database
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: PIMPY
  name: Publicly Available Content Database
  url: http://search.proquest.com/publiccontent
  sourceTypes: Aggregation Database
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
Biology
Physics
EISSN 1560-2281
EndPage 036501
ExternalDocumentID PMC10013193
A845291273
36925620
10_1117_1_JBO_28_3_036501
Genre Research Support, U.S. Gov't, Non-P.H.S
Journal Article
GrantInformation_xml – fundername: National Science Foundation
  grantid: CBET-1554516
– fundername: National Science Foundation funding program named Industry-University Cooperative Research Program Center for Bioanalytical Metrology
  grantid: 1916601
GroupedDBID ---
0R~
29J
4.4
53G
5GY
AAFWJ
ABPTK
ACGFO
ACGFS
ADBBV
AENEX
AFPKN
ALMA_UNASSIGNED_HOLDINGS
BCNDV
CS3
DU5
EBS
F5P
FQ0
GROUPED_DOAJ
HZ~
O9-
OK1
P2P
RNS
RPM
SPBNH
UPT
UT2
W2D
YQT
AAYXX
ACBEA
AEUYN
AFFHD
AFKRA
AKROS
BBNVY
BENPR
BHPHI
CCPQU
CITATION
HCIFZ
HYE
IAO
IGS
INH
M4X
M7P
PBYJJ
PHGZM
PHGZT
PIMPY
PQGLB
CGR
CUY
CVF
ECM
EIF
EJD
EMOBN
M4W
NPM
NU.
7QF
7QO
7QQ
7SC
7SE
7SP
7SR
7TA
7TB
7U5
8BQ
8FD
8FE
8FH
ABUWG
AZQEC
DWQXO
F28
FR3
GNUQQ
H8D
H8G
JG9
JQ2
KR7
L7M
LK8
L~C
L~D
P64
PKEHL
PQEST
PQQKQ
PQUKI
7X8
5PM
ID FETCH-LOGICAL-c534t-c792a981a854bd2ba766af766837e660b8429ef9b484fe6221dd00a3a189f23f3
IEDL.DBID M7P
ISICitedReferencesCount 3
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000961031500012&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1083-3668
1560-2281
IngestDate Tue Nov 04 02:07:18 EST 2025
Sun Nov 09 03:24:09 EST 2025
Fri Jul 25 11:41:11 EDT 2025
Sat Nov 29 13:47:30 EST 2025
Sat Nov 29 10:31:50 EST 2025
Thu Jan 02 22:31:41 EST 2025
Sat Nov 29 05:08:33 EST 2025
Tue Nov 18 21:30:34 EST 2025
Sat Apr 01 05:15:35 EDT 2023
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 3
Keywords generative adversarial networks
fully convolutional networks
small datasets
super-resolution
dense encoder-decoder
biomedical imaging
diffraction-limited
convolutional neural networks
machine learning
fluorescence microscopy
dense layer
Language English
License 2023 The Authors.
Published by SPIE under a Creative Commons Attribution 4.0 International License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c534t-c792a981a854bd2ba766af766837e660b8429ef9b484fe6221dd00a3a189f23f3
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ORCID 0000-0002-1866-6092
OpenAccessLink https://www.proquest.com/docview/2862339285?pq-origsite=%requestingapplication%
PMID 36925620
PQID 2862339285
PQPubID 2049439
PageCount 1
ParticipantIDs pubmed_primary_36925620
crossref_primary_10_1117_1_JBO_28_3_036501
proquest_journals_2862339285
proquest_miscellaneous_2788796121
spie_journals_10_1117_1_JBO_28_3_036501
gale_infotracmisc_A845291273
pubmedcentral_primary_oai_pubmedcentral_nih_gov_10013193
gale_infotracacademiconefile_A845291273
crossref_citationtrail_10_1117_1_JBO_28_3_036501
PublicationCentury 2000
PublicationDate 2023-03-01
PublicationDateYYYYMMDD 2023-03-01
PublicationDate_xml – month: 03
  year: 2023
  text: 2023-03-01
  day: 01
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
– name: Bellingham
PublicationTitle Journal of biomedical optics
PublicationTitleAlternate J. Biomed. Opt
PublicationYear 2023
Publisher Society of Photo-Optical Instrumentation Engineers
SPIE
S P I E - International Society for
Publisher_xml – name: Society of Photo-Optical Instrumentation Engineers
– name: SPIE
– name: S P I E - International Society for
References r4
r5
r6
r7
r8
r9
Glorot (r64) 2011
Ioffe (r65) 2015
r52
r51
r10
r54
r53
r12
r56
r14
r58
r13
r57
r16
r59
r18
r17
Wang (r79) 2018
r19
Mannam (r37) 2020
Shocher (r50) 2018
Goodfellow (r24) 2016
r61
r60
r63
r21
Bevilacqua (r55) 2012
r20
r23
r67
r22
r66
r25
r69
r27
r26
Mannam (r89) 2022
Webb (r2) 2017
Howard (r28) 2020
Yeh (r74) 2019
Jégou (r68) 2017
r70
r72
r30
Mannam (r38) 2020
r73
r32
Kingma (r71) 2014
r76
r31
r75
r34
r78
r77
r36
r35
r39
Hemdan (r85) 2020
Brownlee (r33) 2019
Huang (r15) 2017
Bronzino (r1) 2000
Long (r62) 2015
Soh (r49) 2020
r81
Zhang (r40) 2018
r80
r83
r82
r84
r43
r87
Zhang (r29) 2019
r42
r86
r45
r44
r88
Robinson (r11) 2010
r47
Liang (r3) 2011
r46
r48
Osokin (r41) 2017
References_xml – ident: r36
  doi: 10.1117/12.2578519
– year: 2019
  ident: r74
– ident: r48
  doi: 10.1109/CVPR46437.2021.00988
– year: 2020
  ident: r85
  article-title: CovidX-net: a framework of deep learning classifiers to diagnose Covid-19 in X-ray images
– ident: r51
  doi: 10.1109/TPAMI.2020.2982166
– ident: r66
  doi: 10.1109/TIP.2017.2662206
– year: 2020
  ident: r37
  article-title: Performance analysis of semi-supervised learning in the small-data regime using VAEs
– ident: r45
  doi: 10.1142/S1793545822300166
– ident: r84
  doi: 10.1364/OE.26.033166
– start-page: MW2A–3
  year: 2020
  ident: r38
  article-title: Instant image denoising plugin for ImageJ using convolutional neural networks
– ident: r18
  doi: 10.1073/pnas.1107547108
– ident: r76
  doi: 10.1098/rsta.2021.0110
– ident: r23
  doi: 10.1038/ncomms12471
– ident: r53
  doi: 10.1109/LSP.2021.3098230
– start-page: 11
  year: 2017
  ident: r68
  article-title: The one hundred layers tiramisu: fully convolutional densenets for semantic segmentation
– ident: r69
  doi: 10.1007/978-3-030-66415-2_31
– ident: r43
  doi: 10.1093/bioinformatics/bty241
– ident: r22
  doi: 10.1364/BOE.9.004077
– start-page: 4700
  year: 2017
  ident: r15
  article-title: Densely connected convolutional networks
– ident: r12
  doi: 10.1038/s41598-018-19426-7
– ident: r35
  doi: 10.1007/978-3-030-15712-8_59
– ident: r26
  doi: 10.1117/12.2578574
– ident: r60
  doi: 10.1364/BOE.10.001044
– ident: r19
  doi: 10.1073/pnas.0406877102
– ident: r34
  doi: 10.1109/BigData.2017.8258006
– ident: r75
  doi: 10.3389/fphy.2020.00118
– start-page: 135.1
  year: 2012
  ident: r55
  article-title: Low-complexity single-image super-resolution based on nonnegative neighbor embedding
– year: 2011
  ident: r3
– start-page: 448
  year: 2015
  ident: r65
  article-title: Batch normalization: accelerating deep network training by reducing internal covariate shift
– ident: r83
  doi: 10.1017/S1431927622000782
– ident: r42
  doi: 10.1109/CVPR.2018.00333
– year: 2020
  ident: r28
  article-title: Packet compressed sensing imaging (PCSI): robust image transmission over noisy channels
– ident: r31
  doi: 10.1109/ICCV48922.2021.00663
– year: 2019
  ident: r33
  article-title: Impact of dataset size on deep learning model skill and performance estimates
– ident: r56
  doi: 10.1109/ICCV.2001.937655
– ident: r88
  doi: 10.26599/BDMA.2018.9020001
– ident: r20
  doi: 10.1103/PhysRevLett.99.228105
– year: 2016
  ident: r24
– ident: r86
  doi: 10.1007/s10489-020-01902-1
– year: 2022
  ident: r89
  article-title: Overcoming fundamental limits of three-dimensional in vivo fluorescence imaging using machine learning
– ident: r87
  doi: 10.1038/s41598-019-42294-8
– ident: r63
  doi: 10.1002/mp.12344
– start-page: 3431
  year: 2015
  ident: r62
  article-title: Fully convolutional networks for semantic segmentation
– ident: r14
  doi: 10.1002/mp.13717
– ident: r17
  doi: 10.1007/BF02956173
– start-page: 384
  year: 2010
  ident: r11
  article-title: “New applications of super-resolution in medical imaging
– ident: r39
  doi: 10.1007/978-3-319-24574-4_28
– ident: r57
  doi: 10.1109/CVPRW.2017.150
– start-page: 3516
  year: 2020
  ident: r49
  article-title: Meta-transfer learning for zero-shot super-resolution
– ident: r44
  doi: 10.1038/ncomms13752
– ident: r59
  doi: 10.1038/s41592-018-0239-0
– ident: r67
  doi: 10.1016/j.jcp.2018.04.018
– ident: r52
  doi: 10.1007/s11554-021-01127-6
– ident: r58
  doi: 10.1007/s11042-019-7397-7
– ident: r72
  doi: 10.1364/JOSAA.12.000058
– start-page: 2472
  year: 2018
  ident: r40
  article-title: Residual dense network for image super-resolution
– ident: r47
  doi: 10.1186/s43074-022-00077-x
– year: 2018
  ident: r79
  article-title: ESRGAN: enhanced super-resolution generative adversarial networks
– ident: r30
  doi: 10.1083/jcb.200903097
– ident: r27
  doi: 10.1117/12.2607515
– ident: r46
  doi: 10.1038/s41467-022-30949-6
– ident: r54
  doi: 10.1109/TCI.2020.3031070
– ident: r5
  doi: 10.1046/j.1365-2818.2000.00710.x
– year: 2017
  ident: r2
– start-page: 315
  year: 2011
  ident: r64
  article-title: Deep sparse rectifier neural networks
– ident: r73
  doi: 10.1117/12.2608910
– ident: r13
  doi: 10.1007/s10278-017-0033-z
– ident: r32
  doi: 10.1007/s00330-021-08320-y
– start-page: 3118
  year: 2018
  ident: r50
  article-title: “zero-shot” super-resolution using deep internal learning
– ident: r82
  doi: 10.1038/s41587-021-01092-2
– ident: r9
  doi: 10.1364/OPTICA.5.000458
– ident: r6
  doi: 10.1038/nmeth929
– year: 2000
  ident: r1
– ident: r70
  doi: 10.1364/OPTICA.426870
– ident: r4
  doi: 10.1364/OL.19.000780
– ident: r10
  doi: 10.1364/OPTICA.448287
– ident: r77
  doi: 10.1109/TIP.2003.819861
– ident: r16
  doi: 10.1038/nmeth817
– ident: r21
  doi: 10.1016/j.bpj.2009.08.002
– ident: r78
  doi: 10.1088/1361-6463/ab0261
– start-page: 11710
  year: 2019
  ident: r29
  article-title: A poisson-gaussian denoising dataset with real fluorescence microscopy images
– ident: r80
  doi: 10.1007/s11760-019-01569-3
– ident: r61
  doi: 10.1038/nbt.4106
– ident: r7
  doi: 10.1126/science.1127344
– ident: r8
  doi: 10.1109/TBME.2018.2814538
– start-page: 2233
  year: 2017
  ident: r41
  article-title: GANs for biological image synthesis
– ident: r81
  doi: 10.1038/s41592-018-0216-7
– year: 2014
  ident: r71
  article-title: Adam: a method for stochastic optimization
– ident: r25
  doi: 10.1088/2515-7647/abac1a
SSID ssj0008696
Score 2.4157467
Snippet Machine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed [signal-to-noise...
SignificanceMachine learning (ML) models based on deep convolutional neural networks have been used to significantly increase microscopy resolution, speed...
SourceID pubmedcentral
proquest
gale
pubmed
crossref
spie
SourceType Open Access Repository
Aggregation Database
Index Database
Enrichment Source
Publisher
StartPage 036501
SubjectTerms Animals
Artificial neural networks
Cattle
Coders
Computer architecture
Data interpretation
Datasets
Encoders-Decoders
Endothelial Cells
Fibroblasts
Fluorescence
Fluorescence microscopy
Humans
Image processing
Image Processing, Computer-Assisted - methods
Image resolution
Information retrieval
Localization
Machine Learning
Magnetic Resonance Imaging - methods
Microscopy
Microscopy - methods
Neural networks
Neural Networks, Computer
Pulmonary arteries
Pulmonary artery
Research methodology
Signal to noise ratio
System effectiveness
Training
Title Small training dataset convolutional neural networks for application-specific super-resolution microscopy
URI http://www.dx.doi.org/10.1117/1.JBO.28.3.036501
https://www.ncbi.nlm.nih.gov/pubmed/36925620
https://www.proquest.com/docview/2862339285
https://www.proquest.com/docview/2788796121
https://pubmed.ncbi.nlm.nih.gov/PMC10013193
Volume 28
WOSCitedRecordID wos000961031500012&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: PRVAON
  databaseName: DOAJ Directory of Open Access Journals
  customDbUrl:
  eissn: 1560-2281
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0008696
  issn: 1083-3668
  databaseCode: DOA
  dateStart: 20190101
  isFulltext: true
  titleUrlDefault: https://www.doaj.org/
  providerName: Directory of Open Access Journals
– providerCode: PRVPQU
  databaseName: Biological Science Database
  customDbUrl:
  eissn: 1560-2281
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0008696
  issn: 1083-3668
  databaseCode: M7P
  dateStart: 20190101
  isFulltext: true
  titleUrlDefault: http://search.proquest.com/biologicalscijournals
  providerName: ProQuest
– providerCode: PRVPQU
  databaseName: ProQuest Central
  customDbUrl:
  eissn: 1560-2281
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0008696
  issn: 1083-3668
  databaseCode: BENPR
  dateStart: 20190101
  isFulltext: true
  titleUrlDefault: https://www.proquest.com/central
  providerName: ProQuest
– providerCode: PRVPQU
  databaseName: Publicly Available Content Database
  customDbUrl:
  eissn: 1560-2281
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0008696
  issn: 1083-3668
  databaseCode: PIMPY
  dateStart: 20190101
  isFulltext: true
  titleUrlDefault: http://search.proquest.com/publiccontent
  providerName: ProQuest
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV3db9MwED-xDiR44GPACIwqSEiTkNLVduI4T2hDmwBBqfhSebIcx9UqdW1pWqT999w5btdOYi-8tFJsR3Z8vjv7fv4dwOvMEmtZxWgGTJIK45LCyipRvMxtmueM-Qj-z095r6cGg6IfDtzqAKtc6USvqKuppTPyI46ut0BjrrK3s98JZY2i6GpIobEDu8SSwD10r7_WxEr6_FwM3YxESKlCVJOx_Ih1Pp586XDVER3U4VnICbOyS9e184Z5ug6dbNWzkdswSWcP_ncwD-F-cEbj40Z6HsEtN9mDO016yss9uLdBVojPPVjU1o9h9O3CjMfxKrtETCjT2i1iQrAHSca3ElOm__M48zpG7zjeCJcndMeTcEpxvZy5eYLb_tA2viCQIF2XuXwCP85Ov797n4SUDYnNRLpIbF5wUyhmVJaWFS9NLqUZ4g_ug52U3VKh_XPDokxVOnSSc1ZVKCTCMFUMuRiKp9CaTCfuGcQ2k0KaypaeQVFIlaeyclluKie6tsoi6K4mTNvAZ04DH-tmX5NrpnGONVda6GaOI3izbjJryDxuqnxIUqBpoeN7rQn3FbB3RJmljxXFrBm6fxEcbNXEBWq3i1dCoIOCqPWVBETwal1MLQn0NnHTJdYhpGdBFG8R7Ddit-62kAU6q7wbgdoSyHUFog3fLpmMzj19OPMcSwX265Bk96pP__wUz28ewQu4y9H1a5B5B9BazJfuJdy2fxajet6GnXyg2rB7ctrrf237M4-2X6b4rP_hc__XX0beQVY
linkProvider ProQuest
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMw1V1bb9MwFD4aHQh44DJugQFBAk1CSlfbieM8IDQu08q6MmkDjSfjOK6o1LWlaUH9U_xGzsmlayextz3wkkixY9nOuTn-_B2Al5El1rKM0RcwQSiMCxIrs0DxNLZhHDNW7OB_7cTdrjo5SQ7X4E99FoZglbVNLAx1NrL0j3ybY-gt0Jmr6O34Z0BZo2h3tU6hUYrFvpv_xiVb_qb9Ab_vK853Px6_3wuqrAKBjUQ4DWyccJMoZlQUphlPTSyl6eEFl2pOylaq0ES7XpKGKuw5yTnLMhyHMEwlPS56Atu9AushCXsD1g_bB4ffFrZfySIjGMPAJhDYZLWPyli8zZqf3n1uctUUTfQaUZWFpvaE5_3BkkM8D9Zs5OO-W3KCu7f_t-m7A7eqcNvfKfXjLqy54QZcKxNwzjfg5hIdIz4v4LA2vwf9o1MzGPh1_gyfcLS5m_qE0a90FVslLtDiViDpcx_jf38JEBDQKVZCYvn5bOwmwcTVeu6fEgySDgTN78OXSxn_A2gMR0P3CHwbSSFNZtOCI1JIFYcyc1FsMidaNos8aNUCom3F2E4DH-hy5RZrplGmNFda6FKmPHi9eGVc0pVcVHmLpE6TKcN2ralOZGDviBRM7yjalWcY4HqwuVITTZBdLa6FTlcmMNdnEufBi0UxvUmwvqEbzbAOYVkTIrHz4GEp5otuC5lgOM5bHqgVBVhUIGL01ZJh_0dBkM4KFqkE-7VFunLWp39OxeOLR_Acru8dH3R0p93dfwI3OAa6JQ5xExrTycw9hav217SfT55VBsGH75etRH8BT5WZag
linkToPdf http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMw1V1bb9MwFD4aHSB44LJxCQwIEmgSUtraThznAaHBqCgbpdIAbU8mcRxRqWtL04L61_h1nJNL105ib3vgJZFix7Kdc3P8-TsALwJDrGUpoy8Qe76IrRcZmXqKJ6Hxw5CxYgf_22HY66nj46i_AX_qszAEq6xtYmGo07Ghf-QtjqG3QGeuglZWwSL6-503k58eZZCindY6nUYpIgd28RuXb_nr7j5-65ecd95_effBqzIMeCYQ_swzYcTjSLFYBX6S8iQOpYwzvOCyzUrZThSaa5tFia_8zErOWZrimETMVJRxkQls9wpsYkju8wZs9ruf-idLP6BkkR2MYZDjCWyy2lNlLGyx5se3n5tcNUUTPUhQZaSpveJ537DiHM8DNxv5ZGBXHGLn9v88lXfgVhWGu3ul3tyFDTvagmtlYs7FFtxcoWnE5wVM1uTbMDg6jYdDt86r4RK-Nrczl7D7lQ5jq8QRWtwKhH3u4rrAXQEKeHS6lRBabj6f2Kk3tbX-u6cEj6SDQot78PVSxn8fGqPxyD4E1wRSyDg1ScEdKaQKfZnaIIxTK9omDRxo18KiTcXkTgMf6nJFF2qmUb40V1roUr4ceLV8ZVLSmFxUeZckUJOJw3ZNXJ3UwN4RWZjeU7RbzzDwdWBnrSaaJrNeXAugrkxjrs-kz4Hny2J6k-B-IzueYx3CuEZEbufAg1Lkl90WMsIwnbcdUGvKsKxAhOnrJaPBj4I4nRXsUhH2a5f05qxP_5yKRxeP4BlcR83Rh93ewWO4wTH-LeGJO9CYTef2CVw1v2aDfPq0sg0ufL9sHfoLzsOiKg
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=Small+training+dataset+convolutional+neural+networks+for+application-specific+super-resolution+microscopy&rft.jtitle=Journal+of+biomedical+optics&rft.au=Mannam%2C+Varun&rft.au=Howard%2C+Scott+S&rft.date=2023-03-01&rft.pub=S+P+I+E+-+International+Society+for&rft.issn=1083-3668&rft.eissn=1560-2281&rft.volume=28&rft.issue=3&rft.spage=36501&rft_id=info:doi/10.1117%2F1.JBO.28.3.036501&rft.externalDBID=HAS_PDF_LINK
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1083-3668&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1083-3668&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1083-3668&client=summon