MAPSeg: Unified Unsupervised Domain Adaptation for Heterogeneous Medical Image Segmentation Based on 3D Masked Autoencoding and Pseudo-Labeling
Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a w...
Uložené v:
| Vydané v: | Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) Ročník 2024; s. 5851 - 5862 |
|---|---|
| Hlavní autori: | , , , , , , , , , , , , |
| Médium: | Konferenčný príspevok.. Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
United States
IEEE
01.06.2024
|
| Predmet: | |
| ISSN: | 1063-6919, 1063-6919 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Abstract | Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a unified UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to centralized, federated, and test-time UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/XuzheZ/MAPSeg/. |
|---|---|
| AbstractList | Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a
UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to
,
, and
UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/Xuzhez/MAPSeg/. Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a unified UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to centralized, federated, and test-time UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/Xuzhez/MAPSeg/. Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a unified UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to centralized, federated, and test-time UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/Xuzhez/MAPSeg/.Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a unified UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to centralized, federated, and test-time UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/Xuzhez/MAPSeg/. |
| Author | Angelini, Elsa Li, Ang Wadhwa, Pathik D. Rasmussen, Jerod M. Jackowski, Andrea Parolin OConnor, Thomas G. Zhang, Xuzhe Posner, Jonathan Laine, Andrew F. Wang, Yun Wu, Yuhao Li, Hai Guo, Jia |
| AuthorAffiliation | 6 University of Rochester 8 Emory University 4 University of Maryland, College Park 3 Télécom Paris, LTCI, Institut Polytechnique de Paris 2 Duke University 7 Universidade Federal de São Paulo 1 Columbia University 5 University of California, Irvine |
| AuthorAffiliation_xml | – name: 1 Columbia University – name: 7 Universidade Federal de São Paulo – name: 2 Duke University – name: 6 University of Rochester – name: 8 Emory University – name: 3 Télécom Paris, LTCI, Institut Polytechnique de Paris – name: 4 University of Maryland, College Park – name: 5 University of California, Irvine |
| Author_xml | – sequence: 1 givenname: Xuzhe surname: Zhang fullname: Zhang, Xuzhe organization: Columbia University – sequence: 2 givenname: Yuhao surname: Wu fullname: Wu, Yuhao organization: Duke University – sequence: 3 givenname: Elsa surname: Angelini fullname: Angelini, Elsa organization: Columbia University – sequence: 4 givenname: Ang surname: Li fullname: Li, Ang organization: University of Maryland,College Park – sequence: 5 givenname: Jia surname: Guo fullname: Guo, Jia organization: Columbia University – sequence: 6 givenname: Jerod M. surname: Rasmussen fullname: Rasmussen, Jerod M. organization: University of California,Irvine – sequence: 7 givenname: Thomas G. surname: OConnor fullname: OConnor, Thomas G. organization: University of Rochester – sequence: 8 givenname: Pathik D. surname: Wadhwa fullname: Wadhwa, Pathik D. organization: University of California,Irvine – sequence: 9 givenname: Andrea Parolin surname: Jackowski fullname: Jackowski, Andrea Parolin organization: Universidade Federal de São Paulo – sequence: 10 givenname: Hai surname: Li fullname: Li, Hai organization: Duke University – sequence: 11 givenname: Jonathan surname: Posner fullname: Posner, Jonathan organization: Duke University – sequence: 12 givenname: Andrew F. surname: Laine fullname: Laine, Andrew F. organization: Columbia University – sequence: 13 givenname: Yun surname: Wang fullname: Wang, Yun organization: Duke University |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/39479533$$D View this record in MEDLINE/PubMed |
| BookMark | eNpVkc1uEzEQgA0qoqXkDSrkI5cN_tlZx1xQSIFWSkQElOvKsWcXw64d1ruVeApeGYeGqpxmxvPpG2vmGTkJMSAhF5zNOWf61err9hMIJeVcMFHOGQPQj8hMK72QwCRIxqrH5IyzShaV5vrkQX5KZil9Z4xJwXmlF0_JqdSl0iDlGfm9WW4_Y_ua3gTfeHQ5pmmPw61PubiMvfGBLp3Zj2b0MdAmDvQKRxxiiwHjlOgGnbemo9e9aZFmV4_hCL81B0lO5CXdmPQjF8tpjBhsdD601ARHtwknF4u12WGX356TJ43pEs6O8ZzcvH_3ZXVVrD9-uF4t14WXshwL0QiouG2kcIsKLFegdgqstWpnS4vCAFhpOcoGuanAaaeAccfQggN0Up6TN3fe_bTr0dn858F09X7wvRl-1dH4-v9O8N_qNt7WnIM4rDIbXh4NQ_w5YRrr3ieLXWf-7qWWXIiqLBUc0BcPh91P-XeGDFzcAR4R79v5ggA8A38AdGScJw |
| CODEN | IEEPAD |
| ContentType | Conference Proceeding Journal Article |
| DBID | 6IE 6IH CBEJK RIE RIO NPM 7X8 5PM |
| DOI | 10.1109/CVPR52733.2024.00559 |
| DatabaseName | IEEE Electronic Library (IEL) Conference Proceedings IEEE Proceedings Order Plan (POP) 1998-present by volume IEEE Xplore All Conference Proceedings IEEE/IET Electronic Library (IEL) IEEE Proceedings Order Plans (POP) 1998-present PubMed MEDLINE - Academic PubMed Central (Full Participant titles) |
| DatabaseTitle | PubMed MEDLINE - Academic |
| DatabaseTitleList | PubMed 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: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher – sequence: 3 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Applied Sciences Computer Science |
| EISBN | 9798350353006 |
| EISSN | 1063-6919 |
| EndPage | 5862 |
| ExternalDocumentID | PMC11520032 39479533 10655195 |
| Genre | orig-research Journal Article |
| GrantInformation_xml | – fundername: NSF grantid: CNS-2112562 funderid: 10.13039/100000001 – fundername: NICHD NIH HHS grantid: R00 HD100593 – fundername: NICHD NIH HHS grantid: R00 HD103912 – fundername: NHLBI NIH HHS grantid: R01 HL121270 – fundername: NIMH NIH HHS grantid: R01 MH121070 – fundername: NIEHS NIH HHS grantid: P30 ES001247 |
| GroupedDBID | 6IE 6IH 6IL 6IN AAWTH ABLEC ADZIZ ALMA_UNASSIGNED_HOLDINGS BEFXN BFFAM BGNUA BKEBE BPEOZ CBEJK CHZPO IEGSK IJVOP OCL RIE RIL RIO 23M 29F 29O 6IK ABDPE ACGFS IPLJI M43 NPM RNS 7X8 5PM |
| ID | FETCH-LOGICAL-i334t-2f2561cf32d865c1757b75ccc7bc4ce2a55c3c1e3fe1a65d9d7501d0ec5d5ed33 |
| IEDL.DBID | RIE |
| ISICitedReferencesCount | 10 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001322555906025&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1063-6919 |
| IngestDate | Tue Sep 30 17:02:42 EDT 2025 Mon Jul 21 11:02:15 EDT 2025 Mon Aug 18 01:31:25 EDT 2025 Wed Aug 27 02:00:49 EDT 2025 |
| IsDoiOpenAccess | false |
| IsOpenAccess | true |
| IsPeerReviewed | false |
| IsScholarly | true |
| Language | English |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-i334t-2f2561cf32d865c1757b75ccc7bc4ce2a55c3c1e3fe1a65d9d7501d0ec5d5ed33 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 Co-senior supervising authors. Co-first authors. |
| OpenAccessLink | https://www.ncbi.nlm.nih.gov/pmc/articles/11520032 |
| PMID | 39479533 |
| PQID | 3122644752 |
| PQPubID | 23479 |
| PageCount | 12 |
| ParticipantIDs | pubmedcentral_primary_oai_pubmedcentral_nih_gov_11520032 proquest_miscellaneous_3122644752 ieee_primary_10655195 pubmed_primary_39479533 |
| PublicationCentury | 2000 |
| PublicationDate | 20240601 |
| PublicationDateYYYYMMDD | 2024-06-01 |
| PublicationDate_xml | – month: 6 year: 2024 text: 20240601 day: 1 |
| PublicationDecade | 2020 |
| PublicationPlace | United States |
| PublicationPlace_xml | – name: United States |
| PublicationTitle | Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) |
| PublicationTitleAbbrev | CVPR |
| PublicationTitleAlternate | Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit |
| PublicationYear | 2024 |
| Publisher | IEEE |
| Publisher_xml | – name: IEEE |
| SSID | ssj0003211698 ssj0023720 |
| Score | 2.5578094 |
| Snippet | Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical... |
| SourceID | pubmedcentral proquest pubmed ieee |
| SourceType | Open Access Repository Aggregation Database Index Database Publisher |
| StartPage | 5851 |
| SubjectTerms | Computational modeling Computer vision generalization Image segmentation Magnetic resonance imaging masked image modeling medical image Pediatrics Reviews self-supervised learning Three-dimensional displays unsupervised domain adaptation |
| Title | MAPSeg: Unified Unsupervised Domain Adaptation for Heterogeneous Medical Image Segmentation Based on 3D Masked Autoencoding and Pseudo-Labeling |
| URI | https://ieeexplore.ieee.org/document/10655195 https://www.ncbi.nlm.nih.gov/pubmed/39479533 https://www.proquest.com/docview/3122644752 https://pubmed.ncbi.nlm.nih.gov/PMC11520032 |
| Volume | 2024 |
| WOSCitedRecordID | wos001322555906025&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 | |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Nb9QwEB3RqgdOBbrA8lG5EtfAJo7jhNvSUhWprSKgaG8rx2O3q2qTVbPhb_CXmXHShVblwM2WE8uyx_azZ94zwDskXJ_ZBKPCZj5KK0UpnWFkMuc85hhPKhMem9Dn5_lsVpQDWT1wYZxzIfjMvedk8OVjYzu-KqMZnilWQ9mCLa11T9baXKhIOspkRT7Q4-JJ8eHwR_mV9cUkHQMTFslWrEgaHlF5CE_eD4v8a5853v3PFj6B0R_Gnig3e9FTeOTqZ7A7QEwxTOB2D36dTctv7vKjIKzpueiibrsVrxctZY6apVnUYopm1XvoBUFaccIRMw0Zmmu6VgyeHfFlSSuRoLqWA3upFp8MV0IJeSTOTHtNmWm3blgqk1slTI2ibF2HTXRqqkCEH8HF8efvhyfR8CZDtJAyXUeJJ4wUWy8TzDNlCXzoSitrra5syo-LKWWljZ30LjaZwgIJksQ4cVahcijlc9ium9q9BKEV0umt8D6WOkV28fpcGVRYFSatqnQMI-7g-aqX3Zjf9u0YDm7Hbk5zgR0cJvTAXMZMC061Ssbwoh_Lzd-ySDWH0o4hvzPKmw9YZ_tuSb24CnrbBJo5hC959Y8GvYbHbGB9ANkb2F7fdO4t7Nif60V7s0-2Osv3g63-BlFy7Yk |
| linkProvider | IEEE |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1bb9MwFLZgILGncSmsXI3Ea6CO7TjhrWxMnWirCDa0t8jxsaFCTaql4W_wlzknyQpD8MCbLSeW5etnn_N9h7FXgLg-cTFEmUtCpEqNKZNAZBPvA6QgJqXtgk2Y5TK9uMjygazecWG8953zmX9Nyc6WD7Vr6akMV3iiSQ3lJrullYpFT9faPalIvMwkWToQ5MQke3P0Of9ICmMSL4IxyWRr0iTtwqj8DVH-6Rj520lzcvCfbbzLRr84ezzfnUb32A1f3WcHA8jkwxJuHrAfi2n-yX95yxFtBio6r5p2QztGg5njem1XFZ-C3fQ2eo6gls_IZ6bGqebrtuGDbYefrnEv4ljXeuAvVfydpUowIY_5wjbfMDNttzWJZVKruK2A541voY7mtuyo8CN2fvL-7GgWDVEZopWUahvFAVGScEHGkCbaIfwwpdHOOVM6ReHFtHbSCS-DFzbRkAGCEgET7zRoD1I-ZHtVXflDxo0GvL9lIQhpFJCRN6TagoYys6os1ZiNqIOLTS-8UVz17Zi9vBq7AlcDmThs1wOFFEQMVkbHY_aoH8vd3zJThpxpxyy9Nsq7D0hp-3pJtfraKW4jbCYnvvjxPxr0gt2ZnS3mxfx0-eEJ26fJ1ruTPWV728vWP2O33fftqrl83s3Yny5O7-g |
| 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%3Abook&rft.genre=proceeding&rft.title=Proceedings+%28IEEE+Computer+Society+Conference+on+Computer+Vision+and+Pattern+Recognition.+Online%29&rft.atitle=MAPSeg%3A+Unified+Unsupervised+Domain+Adaptation+for+Heterogeneous+Medical+Image+Segmentation+Based+on+3D+Masked+Autoencoding+and+Pseudo-Labeling&rft.au=Zhang%2C+Xuzhe&rft.au=Wu%2C+Yuhao&rft.au=Angelini%2C+Elsa&rft.au=Li%2C+Ang&rft.date=2024-06-01&rft.pub=IEEE&rft.eissn=1063-6919&rft.spage=5851&rft.epage=5862&rft_id=info:doi/10.1109%2FCVPR52733.2024.00559&rft.externalDocID=10655195 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1063-6919&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1063-6919&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1063-6919&client=summon |