Rethinking Multiple Instance Learning for Whole Slide Image Classification: A Good Instance Classifier Is All You Need

Weakly supervised whole slide image classification is usually formulated as a multiple instance learning (MIL) problem, where each slide is treated as a bag, and the patches cut out of it are treated as instances. Existing methods either train an instance classifier through pseudo-labeling or aggreg...

Celý popis

Uloženo v:

Podrobná bibliografie
Vydáno v:	IEEE transactions on circuits and systems for video technology Ročník 34; číslo 10; s. 9732 - 9744
Hlavní autoři:	Qu, Linhao, Ma, Yingfan, Luo, Xiaoyuan, Guo, Qinhao, Wang, Manning, Song, Zhijian
Médium:	Journal Article
Jazyk:	angličtina
Vydáno:	New York IEEE 01.10.2024 The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:	Aggregates Algorithms Attention Circuits and systems Classification Contrastive learning Feature extraction Image classification Labels Machine learning Multiple instance learning Noise measurement prototype learning Prototypes Training whole slide image classification
ISSN:	1051-8215, 1558-2205
On-line přístup:	Získat plný text
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Abstract	Weakly supervised whole slide image classification is usually formulated as a multiple instance learning (MIL) problem, where each slide is treated as a bag, and the patches cut out of it are treated as instances. Existing methods either train an instance classifier through pseudo-labeling or aggregate instance features into a bag feature through attention mechanisms and then train a bag classifier, where the attention scores can be used for instance-level classification. However, the pseudo instance labels constructed by the former usually contain a lot of noise, and the attention scores constructed by the latter are not accurate enough, both of which affect their performance. In this paper, we propose an instance-level MIL framework based on contrastive learning and prototype learning to effectively accomplish both instance classification and bag classification tasks. To this end, we propose an instance-level weakly supervised contrastive learning algorithm for the first time under the MIL setting to effectively learn instance feature representation. We also propose an accurate pseudo label generation method through prototype learning. We then develop a joint training strategy for weakly supervised contrastive learning, prototype learning, and instance classifier training. Extensive experiments and visualizations on four datasets demonstrate the powerful performance of our method. Codes are available at https://github.com/miccaiif/INS .
AbstractList	Weakly supervised whole slide image classification is usually formulated as a multiple instance learning (MIL) problem, where each slide is treated as a bag, and the patches cut out of it are treated as instances. Existing methods either train an instance classifier through pseudo-labeling or aggregate instance features into a bag feature through attention mechanisms and then train a bag classifier, where the attention scores can be used for instance-level classification. However, the pseudo instance labels constructed by the former usually contain a lot of noise, and the attention scores constructed by the latter are not accurate enough, both of which affect their performance. In this paper, we propose an instance-level MIL framework based on contrastive learning and prototype learning to effectively accomplish both instance classification and bag classification tasks. To this end, we propose an instance-level weakly supervised contrastive learning algorithm for the first time under the MIL setting to effectively learn instance feature representation. We also propose an accurate pseudo label generation method through prototype learning. We then develop a joint training strategy for weakly supervised contrastive learning, prototype learning, and instance classifier training. Extensive experiments and visualizations on four datasets demonstrate the powerful performance of our method. Codes are available at https://github.com/miccaiif/INS .
Author	Song, Zhijian Luo, Xiaoyuan Guo, Qinhao Ma, Yingfan Qu, Linhao Wang, Manning
Author_xml	– sequence: 1 givenname: Linhao orcidid: 0000-0001-8815-7050 surname: Qu fullname: Qu, Linhao email: lhqu20@fudan.edu.cn organization: Digital Medical Research Center, School of Basic Medical Science, Fudan University, Shanghai, China – sequence: 2 givenname: Yingfan surname: Ma fullname: Ma, Yingfan email: 22211010089@m.fudan.edu.cn organization: Digital Medical Research Center, School of Basic Medical Science, Fudan University, Shanghai, China – sequence: 3 givenname: Xiaoyuan orcidid: 0000-0002-8456-5847 surname: Luo fullname: Luo, Xiaoyuan email: 19111010030@fudan.edu.cn organization: Digital Medical Research Center, School of Basic Medical Science, Fudan University, Shanghai, China – sequence: 4 givenname: Qinhao surname: Guo fullname: Guo, Qinhao email: 18111230017@fudan.edu.cn organization: Department of Gynecologic Oncology, Fudan University Shanghai Cancer Center, Shanghai, China – sequence: 5 givenname: Manning orcidid: 0000-0002-9255-3897 surname: Wang fullname: Wang, Manning email: mnwang@fudan.edu.cn organization: Digital Medical Research Center, School of Basic Medical Science, Fudan University, Shanghai, China – sequence: 6 givenname: Zhijian orcidid: 0000-0001-9873-216X surname: Song fullname: Song, Zhijian email: zjsong@fudan.edu.cn organization: Digital Medical Research Center, School of Basic Medical Science, Fudan University, Shanghai, China
BookMark	eNp9kU9LAzEQxYMoqNUvIB4Cnrcm2WQ3660UrYWqYKviacnuTjS6JjXZCn570z-iePA0A-_9Zpg3-2jbOgsIHVHSp5QUp7Ph9H7WZ4TxfsoJkXm2hfaoEDJhjIjt2BNBE8mo2EX7IbwQQrnk-R76uIXu2dhXY5_w1aLtzLwFPLahU7YGPAHl7VLSzuOHZxe1aWua6HhTT4CHrQrBaFOrzjh7hgd45Fzzg3_r4PE44EHb4ke3wNcAzQHa0aoNcLipPXR3cT4bXiaTm9F4OJgkNSuyLhFFxTnjoskF1ULJrE41ZZkgGYWGNnWhCGSSUy25ziRjVcUISWWEa5nrSqY9dLKeO_fufQGhK1_cwtu4skwpY3GJLIrokmtX7V0IHnRZm251U-eVaUtKymXK5SrlcplyuUk5ouwPOvfmTfnP_6HjNWQA4Bcg0viWIv0CebeKHA
CODEN	ITCTEM
CitedBy_id	crossref_primary_10_1016_j_modpat_2024_100648 crossref_primary_10_1016_j_prp_2025_156006 crossref_primary_10_1016_j_compbiomed_2025_110075 crossref_primary_10_1016_j_cmpb_2024_108491 crossref_primary_10_1109_ACCESS_2025_3578651 crossref_primary_10_1016_j_jiixd_2025_05_001 crossref_primary_10_1007_s10489_025_06300_z crossref_primary_10_1109_TCSVT_2025_3528625 crossref_primary_10_1007_s13369_025_10320_9 crossref_primary_10_1016_j_media_2025_103631 crossref_primary_10_1038_s41746_025_01580_8 crossref_primary_10_1016_j_media_2025_103647 crossref_primary_10_3390_electronics13224445 crossref_primary_10_1109_TMI_2025_3564976 crossref_primary_10_1016_j_compbiomed_2025_110634 crossref_primary_10_3390_math13132178
Cites_doi	10.1109/TMI.2022.3227066 10.1088/1361-6560/ac910a 10.1109/TCSVT.2023.3294938 10.1007/978-3-030-87237-3_32 10.1109/TMI.2022.3176598 10.1038/s41591-019-0508-1 10.1109/CVPR42600.2020.00975 10.1109/CVPR46437.2021.01409 10.1007/978-3-031-16434-7_3 10.1109/CVPR52688.2022.01567 10.59275/j.melba.2023-5g54 10.1109/TMI.2023.3241204 10.1109/tmi.2023.3253760 10.1109/TMI.2020.3021387 10.1038/s41586-021-03512-4 10.1007/978-3-031-26387-3_26 10.1609/aaai.v34i04.6030 10.1109/TCSVT.2021.3128054 10.1016/j.media.2023.102748 10.1093/bioinformatics/btw252 10.1002/path.6027 10.1038/s41598-020-66333-x 10.1109/ICCV48922.2021.00398 10.1109/TCSVT.2022.3219893 10.1109/TCSVT.2022.3141051 10.1109/TCSVT.2022.3169469 10.1016/j.patcog.2017.08.026 10.1109/CVPR52688.2022.01824 10.1109/TMI.2019.2927182 10.1016/j.media.2020.101789 10.1109/CVPR42600.2020.00391 10.1109/TCSVT.2023.3281151 10.1016/j.modpat.2023.100316 10.1109/TCSVT.2020.2992276 10.1016/j.media.2019.03.009 10.1007/978-3-030-59722-1_50 10.1109/ICCV48922.2021.00951 10.1109/TCSVT.2022.3209209 10.1038/s41551-020-00682-w 10.1109/CVPR.2009.5206848 10.5555/3524938.3525087 10.1016/j.media.2021.102298 10.1016/j.media.2020.101813 10.1109/TCSVT.2022.3207174
ContentType	Journal Article
Copyright	Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024
Copyright_xml	– notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024
DBID	97E RIA RIE AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D
DOI	10.1109/TCSVT.2024.3400876
DatabaseName	IEEE All-Society Periodicals Package (ASPP) 2005–Present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE/IET Electronic Library CrossRef Computer and Information Systems Abstracts Electronics & Communications Abstracts Technology Research Database ProQuest Computer Science Collection Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Academic Computer and Information Systems Abstracts Professional
DatabaseTitle	CrossRef Technology Research Database Computer and Information Systems Abstracts – Academic Electronics & Communications Abstracts ProQuest Computer Science Collection Computer and Information Systems Abstracts Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Professional
DatabaseTitleList	Technology Research Database
Database_xml	– sequence: 1 dbid: RIE name: IEEE Xplore url: https://ieeexplore.ieee.org/ sourceTypes: Publisher
DeliveryMethod	fulltext_linktorsrc
Discipline	Engineering
EISSN	1558-2205
EndPage	9744
ExternalDocumentID	10_1109_TCSVT_2024_3400876 10530149
Genre	orig-research
GrantInformation_xml	– fundername: National Natural Science Foundation of China grantid: 82072021 funderid: 10.13039/501100001809
GroupedDBID	-~X 0R~ 29I 4.4 5GY 5VS 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABQJQ ABVLG ACGFO ACGFS ACIWK AENEX AETIX AGQYO AGSQL AHBIQ AI. AIBXA AKJIK AKQYR ALLEH ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 EBS EJD HZ~ H~9 ICLAB IFIPE IFJZH IPLJI JAVBF LAI M43 O9- OCL P2P RIA RIE RNS RXW TAE TN5 VH1 AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D
ID	FETCH-LOGICAL-c296t-59b44245d751f5a86c3f1265061ed1dc9a0e6841f84f6822bb20038c29c87fb83
IEDL.DBID	RIE
ISICitedReferencesCount	17
ISICitedReferencesURI	http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001346503100074&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN	1051-8215
IngestDate	Mon Jun 30 10:20:29 EDT 2025 Sat Nov 29 01:44:28 EST 2025 Tue Nov 18 22:38:11 EST 2025 Wed Aug 27 01:58:40 EDT 2025
IsPeerReviewed	true
IsScholarly	true
Issue	10
Language	English
License	https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c296t-59b44245d751f5a86c3f1265061ed1dc9a0e6841f84f6822bb20038c29c87fb83
Notes	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14
ORCID	0000-0002-8456-5847 0000-0002-9255-3897 0000-0001-9873-216X 0000-0001-8815-7050
PQID	3122296899
PQPubID	85433
PageCount	13
ParticipantIDs	crossref_citationtrail_10_1109_TCSVT_2024_3400876 proquest_journals_3122296899 ieee_primary_10530149 crossref_primary_10_1109_TCSVT_2024_3400876
PublicationCentury	2000
PublicationDate	2024-10-01
PublicationDateYYYYMMDD	2024-10-01
PublicationDate_xml	– month: 10 year: 2024 text: 2024-10-01 day: 01
PublicationDecade	2020
PublicationPlace	New York
PublicationPlace_xml	– name: New York
PublicationTitle	IEEE transactions on circuits and systems for video technology
PublicationTitleAbbrev	TCSVT
PublicationYear	2024
Publisher	IEEE The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Publisher_xml	– name: IEEE – name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
References	ref13 ref12 ref15 ref52 ref55 ref10 ref17 ref16 ref19 ref51 ref50 ref46 ref45 ref48 Yang (ref53) 2023 ref47 ref42 ref41 ref44 ref43 Lu (ref11) 2019 ref49 ref8 Shao (ref24); 34 ref9 Qu (ref14); 35 ref4 ref3 ref6 ref5 Qu (ref7); 36 ref40 ref35 ref34 Chen (ref38) 2022 ref37 ref36 ref30 ref33 ref32 ref2 Ilse (ref18) ref1 ref39 Wang (ref54) 2022 Yan (ref28) Maron (ref31); 10 ref23 ref26 ref25 ref20 ref22 Wang (ref27); 35 ref21 ref29
References_xml	– ident: ref33 doi: 10.1109/TMI.2022.3227066 – ident: ref1 doi: 10.1088/1361-6560/ac910a – ident: ref6 doi: 10.1109/TCSVT.2023.3294938 – start-page: 2127 volume-title: Proc. Int. Conf. Mach. Learn. ident: ref18 article-title: Attention-based deep multiple instance learning – ident: ref26 doi: 10.1007/978-3-030-87237-3_32 – ident: ref34 doi: 10.1109/TMI.2022.3176598 – ident: ref15 doi: 10.1038/s41591-019-0508-1 – volume: 35 start-page: 18009 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref27 article-title: SCL-WC: Cross-slide contrastive learning for weakly-supervised whole-slide image classification – ident: ref41 doi: 10.1109/CVPR42600.2020.00975 – ident: ref22 doi: 10.1109/CVPR46437.2021.01409 – ident: ref13 doi: 10.1007/978-3-031-16434-7_3 – ident: ref37 doi: 10.1109/CVPR52688.2022.01567 – ident: ref8 doi: 10.59275/j.melba.2023-5g54 – ident: ref36 doi: 10.1109/TMI.2023.3241204 – ident: ref35 doi: 10.1109/tmi.2023.3253760 – volume: 10 start-page: 1 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref31 article-title: A framework for multiple-instance learning – ident: ref10 doi: 10.1109/TMI.2020.3021387 – ident: ref3 doi: 10.1038/s41586-021-03512-4 – volume: 34 start-page: 2136 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref24 article-title: TransMIL: Transformer based correlated multiple instance learning for whole slide image classification – start-page: 662 volume-title: Proc. Asian Conf. Mach. Learn. (ACML) ident: ref28 article-title: Deep multi-instance learning with dynamic pooling – year: 2022 ident: ref38 article-title: Self-supervised vision transformers learn visual concepts in histopathology publication-title: arXiv:2203.00585 – ident: ref51 doi: 10.1007/978-3-031-26387-3_26 – ident: ref21 doi: 10.1609/aaai.v34i04.6030 – ident: ref48 doi: 10.1109/TCSVT.2021.3128054 – ident: ref52 doi: 10.1016/j.media.2023.102748 – ident: ref30 doi: 10.1093/bioinformatics/btw252 – ident: ref39 doi: 10.1002/path.6027 – ident: ref17 doi: 10.1038/s41598-020-66333-x – ident: ref25 doi: 10.1109/ICCV48922.2021.00398 – ident: ref44 doi: 10.1109/TCSVT.2022.3219893 – ident: ref46 doi: 10.1109/TCSVT.2022.3141051 – year: 2023 ident: ref53 article-title: TPMIL: Trainable prototype enhanced multiple instance learning for whole slide image classification publication-title: arXiv:2305.00696 – ident: ref45 doi: 10.1109/TCSVT.2022.3169469 – year: 2019 ident: ref11 article-title: Semi-supervised histology classification using deep multiple instance learning and contrastive predictive coding publication-title: arXiv:1910.10825 – ident: ref29 doi: 10.1016/j.patcog.2017.08.026 – ident: ref23 doi: 10.1109/CVPR52688.2022.01824 – year: 2022 ident: ref54 article-title: PiCO+: Contrastive label disambiguation for robust partial label learning publication-title: arXiv:2201.08984 – ident: ref4 doi: 10.1109/TMI.2019.2927182 – ident: ref20 doi: 10.1016/j.media.2020.101789 – ident: ref19 doi: 10.1109/CVPR42600.2020.00391 – volume: 36 start-page: 1 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref7 article-title: The rise of ai language pathologists: Exploring two-level prompt learning for few-shot weakly-supervised whole slide image classification – ident: ref50 doi: 10.1109/TCSVT.2023.3281151 – ident: ref32 doi: 10.1016/j.modpat.2023.100316 – ident: ref47 doi: 10.1109/TCSVT.2020.2992276 – ident: ref9 doi: 10.1016/j.media.2019.03.009 – ident: ref16 doi: 10.1007/978-3-030-59722-1_50 – ident: ref42 doi: 10.1109/ICCV48922.2021.00951 – ident: ref49 doi: 10.1109/TCSVT.2022.3209209 – ident: ref12 doi: 10.1038/s41551-020-00682-w – ident: ref55 doi: 10.1109/CVPR.2009.5206848 – ident: ref40 doi: 10.5555/3524938.3525087 – volume: 35 start-page: 15368 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref14 article-title: Bi-directional weakly supervised knowledge distillation for whole slide image classification – ident: ref5 doi: 10.1016/j.media.2021.102298 – ident: ref2 doi: 10.1016/j.media.2020.101813 – ident: ref43 doi: 10.1109/TCSVT.2022.3207174
SSID	ssj0014847
Score	2.55881
Snippet	Weakly supervised whole slide image classification is usually formulated as a multiple instance learning (MIL) problem, where each slide is treated as a bag,...
SourceID	proquest crossref ieee
SourceType	Aggregation Database Enrichment Source Index Database Publisher
StartPage	9732
SubjectTerms	Aggregates Algorithms Attention Circuits and systems Classification Contrastive learning Feature extraction Image classification Labels Machine learning Multiple instance learning Noise measurement prototype learning Prototypes Training whole slide image classification
Title	Rethinking Multiple Instance Learning for Whole Slide Image Classification: A Good Instance Classifier Is All You Need
URI	https://ieeexplore.ieee.org/document/10530149 https://www.proquest.com/docview/3122296899
Volume	34
WOSCitedRecordID	wos001346503100074&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: PRVIEE databaseName: IEEE Xplore customDbUrl: eissn: 1558-2205 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0014847 issn: 1051-8215 databaseCode: RIE dateStart: 19910101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NT8IwFG-UeNCDnxhRND14M8OVtV3rjRBREiVGULktW_emJBMMDP5-225DjNHE25K9tyz7tX3v7X38EDpPhEiUjhscIShzKKjQiUJGHR4ClUIpj8Z2ZP6d3-uJ4VA-FM3qthcGAGzxGTTMpc3lxxM1N7_K9A5nJgKQ62jd93nerLVMGVBh2cS0DHGENmRlh4wrLwft_vNAx4JN2vCoHcL2zQpZWpUfZ7E1MJ2df77aLtouPEncyqHfQ2sw3kdbK_MFD9DiEbK3nBwB3xeVg7hrHUIFuBit-oq134pfDE8u7qejWEu860MGW7pMU0hksbvCLXwzmcRf6uV9mOLuDLfSFOuTA_e0Nayip871oH3rFEQLjmpKnjlMRtRkQGOfkYSFgisvIU3tu3ECMYmVDF3ggpJE0IRrjyKKTEmb0MpK-EkkvENUGU_GcIQwQDPioRu7wEIqiQq1ywO-GzGPR8CoqiFSfvhAFVPIDRlGGthoxJWBBSswYAUFWDV0sdT5yGdw_CldNfCsSObI1FC9BDgo9uks8IjhM-c66Dz-Re0EbZqn5_V7dVTJpnM4RRtqkY1m0zO7BD8BktnX9g
linkProvider	IEEE
linkToHtml	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3JTsMwEB2xScCBHVFWH7ihlLixXZtbhVgq2gpBgd6ixJkAUmlRW_h-bMctIAQSt0iZUaI82zOTWR7AYS5lrk3cEEjJeMBQJ0GacBaIBJmSWkcscyPzG9VWS3Y66to3q7teGER0xWdYtpcul5_19Zv9VWZ2OLcRgJqGWUud5du1JkkDJh2fmJGigTSmbNwjE6rj9untfdtEgxVWjpgbw_bNDjlilR-nsTMx58v_fLkVWPK-JKkV4K_CFPbWYPHLhMF1eL_B0VNBj0CavnaQ1J1LqJH44aqPxHiu5MEy5ZLb7nNmJF7MMUMcYaYtJXLonZAauej3s0_18X0ckPqQ1LpdYs4O0jL2cAPuzs_ap5eBp1oIdEWJUcBVymwONKtymvNECh3ltGK8N0Exo5lWSYhCMppLlgvjU6SpLWqTRlnLap7KaBNmev0ebgFBrKQiCbMQecIU1YlxerAapjwSKXKmS0DHHz7Wfg65pcPoxi4eCVXswIotWLEHqwRHE53XYgrHn9IbFp4vkgUyJdgdAxz7nTqMI2oZzYUJO7d_UTuA-ct2sxE36q2rHViwTyqq-XZhZjR4wz2Y0--j5-Fg3y3HD_0A2z8
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=Rethinking+Multiple+Instance+Learning+for+Whole+Slide+Image+Classification%3A+A+Good+Instance+Classifier+Is+All+You+Need&rft.jtitle=IEEE+transactions+on+circuits+and+systems+for+video+technology&rft.au=Qu%2C+Linhao&rft.au=Ma%2C+Yingfan&rft.au=Luo%2C+Xiaoyuan&rft.au=Guo%2C+Qinhao&rft.date=2024-10-01&rft.pub=IEEE&rft.issn=1051-8215&rft.volume=34&rft.issue=10&rft.spage=9732&rft.epage=9744&rft_id=info:doi/10.1109%2FTCSVT.2024.3400876&rft.externalDocID=10530149
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1051-8215&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1051-8215&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1051-8215&client=summon