Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images

Purpose A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data f...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:International journal for computer assisted radiology and surgery 2024-11, Vol.19 (11), p.2153-2163
Hauptverfasser: Ichiuji, Yoshihito, Mabu, Shingo, Hatta, Satomi, Inai, Kunihiro, Higuchi, Shohei, Kido, Shoji
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 2163
container_issue 11
container_start_page 2153
container_title International journal for computer assisted radiology and surgery
container_volume 19
creator Ichiuji, Yoshihito
Mabu, Shingo
Hatta, Satomi
Inai, Kunihiro
Higuchi, Shohei
Kido, Shoji
description Purpose A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data for deep learning. In that case, we can consider a case where a classifier trained at a certain medical institution that has a sufficient number of data is reused at other institutions. However, when using data from multiple institutions, it is necessary to unify the feature distribution because the feature of the data differs due to differences in data acquisition conditions. Methods To unify the feature distribution, the data from Institution T are transformed to have the closer distribution to that from Institution S by applying a domain transformation using semi-supervised CycleGAN. The proposed method enhances CycleGAN considering the feature distribution of classes for making appropriate domain transformation for classification. In addition, to address the problem of imbalanced data with different numbers of data for each cancer type, several methods dealing with imbalanced data are applied to semi-supervised CycleGAN. Results The experimental results showed that the classification performance was enhanced when the dataset from Institution S was used as training data and the testing dataset from Institution T was classified after applying domain transformation. In addition, focal loss contributed to improving the mean F1 score the best as a method that addresses the class imbalance. Conclusion The proposed method achieved the domain transformation of thyroid tissue images between two domains, where it retained the important features related to the classes across domains and showed the best F1 score with significant differences compared with other methods. In addition, the proposed method was further enhanced by addressing the class imbalance of the dataset.
doi_str_mv 10.1007/s11548-024-03061-x
format Article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_2929058934</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2929058934</sourcerecordid><originalsourceid>FETCH-LOGICAL-c298t-e421e413624f19ae284357a7b0c7d760c9319a48ef6ae91d02d885353026f53e3</originalsourceid><addsrcrecordid>eNp9kEtPwzAQhC0EgvL4AxyQj1wCfiWxj6g8pQoucLZMsilGSVy8SUX_PS4tiBMnrzzfjHaHkFPOLjhj5SVyniudMaEyJlnBs88dMuE6DYUSZvfPfEAOEd8ZU3kp831yILWQWhkxIXgdOud7OkTXYxNi5wYfejqi7-cUofMZjguIS49Q0-mqauHu6pEmkPpuEcNyjSX929lXQENDq9Yh-ma1loa3VQy-poNHHCF53BzwmOw1rkU42b5H5OX25nl6n82e7h6mV7OsEkYPGSjBQXFZCNVw40BoJfPSla-sKuuyYJWR6VtpaAoHhtdM1FrnMpdMFE0uQR6R801uWvRjBBxs57GCtnU9hBGtMMKwXBupEio2aBUDYoTGLmJaNq4sZ3Zdtt2UbVPZ9rts-5lMZ9v88bWD-tfy024C5AbAJPVziPY9jLFPN_8X-wUqz4y0</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2929058934</pqid></control><display><type>article</type><title>Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images</title><source>MEDLINE</source><source>SpringerLink Journals - AutoHoldings</source><creator>Ichiuji, Yoshihito ; Mabu, Shingo ; Hatta, Satomi ; Inai, Kunihiro ; Higuchi, Shohei ; Kido, Shoji</creator><creatorcontrib>Ichiuji, Yoshihito ; Mabu, Shingo ; Hatta, Satomi ; Inai, Kunihiro ; Higuchi, Shohei ; Kido, Shoji</creatorcontrib><description>Purpose A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data for deep learning. In that case, we can consider a case where a classifier trained at a certain medical institution that has a sufficient number of data is reused at other institutions. However, when using data from multiple institutions, it is necessary to unify the feature distribution because the feature of the data differs due to differences in data acquisition conditions. Methods To unify the feature distribution, the data from Institution T are transformed to have the closer distribution to that from Institution S by applying a domain transformation using semi-supervised CycleGAN. The proposed method enhances CycleGAN considering the feature distribution of classes for making appropriate domain transformation for classification. In addition, to address the problem of imbalanced data with different numbers of data for each cancer type, several methods dealing with imbalanced data are applied to semi-supervised CycleGAN. Results The experimental results showed that the classification performance was enhanced when the dataset from Institution S was used as training data and the testing dataset from Institution T was classified after applying domain transformation. In addition, focal loss contributed to improving the mean F1 score the best as a method that addresses the class imbalance. Conclusion The proposed method achieved the domain transformation of thyroid tissue images between two domains, where it retained the important features related to the classes across domains and showed the best F1 score with significant differences compared with other methods. In addition, the proposed method was further enhanced by addressing the class imbalance of the dataset.</description><identifier>ISSN: 1861-6429</identifier><identifier>EISSN: 1861-6429</identifier><identifier>DOI: 10.1007/s11548-024-03061-x</identifier><identifier>PMID: 38238492</identifier><language>eng</language><publisher>Cham: Springer International Publishing</publisher><subject>Computer Imaging ; Computer Science ; Deep Learning ; Health Informatics ; Humans ; Image Interpretation, Computer-Assisted - methods ; Imaging ; Medicine ; Medicine &amp; Public Health ; Original Article ; Pattern Recognition and Graphics ; Radiology ; Surgery ; Thyroid Gland - diagnostic imaging ; Thyroid Neoplasms - diagnostic imaging ; Thyroid Neoplasms - pathology ; Vision</subject><ispartof>International journal for computer assisted radiology and surgery, 2024-11, Vol.19 (11), p.2153-2163</ispartof><rights>CARS 2024. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.</rights><rights>2024. CARS.</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c298t-e421e413624f19ae284357a7b0c7d760c9319a48ef6ae91d02d885353026f53e3</cites><orcidid>0000-0002-8759-8337</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://link.springer.com/content/pdf/10.1007/s11548-024-03061-x$$EPDF$$P50$$Gspringer$$H</linktopdf><linktohtml>$$Uhttps://link.springer.com/10.1007/s11548-024-03061-x$$EHTML$$P50$$Gspringer$$H</linktohtml><link.rule.ids>314,780,784,27924,27925,41488,42557,51319</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/38238492$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Ichiuji, Yoshihito</creatorcontrib><creatorcontrib>Mabu, Shingo</creatorcontrib><creatorcontrib>Hatta, Satomi</creatorcontrib><creatorcontrib>Inai, Kunihiro</creatorcontrib><creatorcontrib>Higuchi, Shohei</creatorcontrib><creatorcontrib>Kido, Shoji</creatorcontrib><title>Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images</title><title>International journal for computer assisted radiology and surgery</title><addtitle>Int J CARS</addtitle><addtitle>Int J Comput Assist Radiol Surg</addtitle><description>Purpose A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data for deep learning. In that case, we can consider a case where a classifier trained at a certain medical institution that has a sufficient number of data is reused at other institutions. However, when using data from multiple institutions, it is necessary to unify the feature distribution because the feature of the data differs due to differences in data acquisition conditions. Methods To unify the feature distribution, the data from Institution T are transformed to have the closer distribution to that from Institution S by applying a domain transformation using semi-supervised CycleGAN. The proposed method enhances CycleGAN considering the feature distribution of classes for making appropriate domain transformation for classification. In addition, to address the problem of imbalanced data with different numbers of data for each cancer type, several methods dealing with imbalanced data are applied to semi-supervised CycleGAN. Results The experimental results showed that the classification performance was enhanced when the dataset from Institution S was used as training data and the testing dataset from Institution T was classified after applying domain transformation. In addition, focal loss contributed to improving the mean F1 score the best as a method that addresses the class imbalance. Conclusion The proposed method achieved the domain transformation of thyroid tissue images between two domains, where it retained the important features related to the classes across domains and showed the best F1 score with significant differences compared with other methods. In addition, the proposed method was further enhanced by addressing the class imbalance of the dataset.</description><subject>Computer Imaging</subject><subject>Computer Science</subject><subject>Deep Learning</subject><subject>Health Informatics</subject><subject>Humans</subject><subject>Image Interpretation, Computer-Assisted - methods</subject><subject>Imaging</subject><subject>Medicine</subject><subject>Medicine &amp; Public Health</subject><subject>Original Article</subject><subject>Pattern Recognition and Graphics</subject><subject>Radiology</subject><subject>Surgery</subject><subject>Thyroid Gland - diagnostic imaging</subject><subject>Thyroid Neoplasms - diagnostic imaging</subject><subject>Thyroid Neoplasms - pathology</subject><subject>Vision</subject><issn>1861-6429</issn><issn>1861-6429</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>EIF</sourceid><recordid>eNp9kEtPwzAQhC0EgvL4AxyQj1wCfiWxj6g8pQoucLZMsilGSVy8SUX_PS4tiBMnrzzfjHaHkFPOLjhj5SVyniudMaEyJlnBs88dMuE6DYUSZvfPfEAOEd8ZU3kp831yILWQWhkxIXgdOud7OkTXYxNi5wYfejqi7-cUofMZjguIS49Q0-mqauHu6pEmkPpuEcNyjSX929lXQENDq9Yh-ma1loa3VQy-poNHHCF53BzwmOw1rkU42b5H5OX25nl6n82e7h6mV7OsEkYPGSjBQXFZCNVw40BoJfPSla-sKuuyYJWR6VtpaAoHhtdM1FrnMpdMFE0uQR6R801uWvRjBBxs57GCtnU9hBGtMMKwXBupEio2aBUDYoTGLmJaNq4sZ3Zdtt2UbVPZ9rts-5lMZ9v88bWD-tfy024C5AbAJPVziPY9jLFPN_8X-wUqz4y0</recordid><startdate>20241101</startdate><enddate>20241101</enddate><creator>Ichiuji, Yoshihito</creator><creator>Mabu, Shingo</creator><creator>Hatta, Satomi</creator><creator>Inai, Kunihiro</creator><creator>Higuchi, Shohei</creator><creator>Kido, Shoji</creator><general>Springer International Publishing</general><scope>CGR</scope><scope>CUY</scope><scope>CVF</scope><scope>ECM</scope><scope>EIF</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0002-8759-8337</orcidid></search><sort><creationdate>20241101</creationdate><title>Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images</title><author>Ichiuji, Yoshihito ; Mabu, Shingo ; Hatta, Satomi ; Inai, Kunihiro ; Higuchi, Shohei ; Kido, Shoji</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c298t-e421e413624f19ae284357a7b0c7d760c9319a48ef6ae91d02d885353026f53e3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Imaging</topic><topic>Computer Science</topic><topic>Deep Learning</topic><topic>Health Informatics</topic><topic>Humans</topic><topic>Image Interpretation, Computer-Assisted - methods</topic><topic>Imaging</topic><topic>Medicine</topic><topic>Medicine &amp; Public Health</topic><topic>Original Article</topic><topic>Pattern Recognition and Graphics</topic><topic>Radiology</topic><topic>Surgery</topic><topic>Thyroid Gland - diagnostic imaging</topic><topic>Thyroid Neoplasms - diagnostic imaging</topic><topic>Thyroid Neoplasms - pathology</topic><topic>Vision</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Ichiuji, Yoshihito</creatorcontrib><creatorcontrib>Mabu, Shingo</creatorcontrib><creatorcontrib>Hatta, Satomi</creatorcontrib><creatorcontrib>Inai, Kunihiro</creatorcontrib><creatorcontrib>Higuchi, Shohei</creatorcontrib><creatorcontrib>Kido, Shoji</creatorcontrib><collection>Medline</collection><collection>MEDLINE</collection><collection>MEDLINE (Ovid)</collection><collection>MEDLINE</collection><collection>MEDLINE</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>MEDLINE - Academic</collection><jtitle>International journal for computer assisted radiology and surgery</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Ichiuji, Yoshihito</au><au>Mabu, Shingo</au><au>Hatta, Satomi</au><au>Inai, Kunihiro</au><au>Higuchi, Shohei</au><au>Kido, Shoji</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images</atitle><jtitle>International journal for computer assisted radiology and surgery</jtitle><stitle>Int J CARS</stitle><addtitle>Int J Comput Assist Radiol Surg</addtitle><date>2024-11-01</date><risdate>2024</risdate><volume>19</volume><issue>11</issue><spage>2153</spage><epage>2163</epage><pages>2153-2163</pages><issn>1861-6429</issn><eissn>1861-6429</eissn><abstract>Purpose A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data for deep learning. In that case, we can consider a case where a classifier trained at a certain medical institution that has a sufficient number of data is reused at other institutions. However, when using data from multiple institutions, it is necessary to unify the feature distribution because the feature of the data differs due to differences in data acquisition conditions. Methods To unify the feature distribution, the data from Institution T are transformed to have the closer distribution to that from Institution S by applying a domain transformation using semi-supervised CycleGAN. The proposed method enhances CycleGAN considering the feature distribution of classes for making appropriate domain transformation for classification. In addition, to address the problem of imbalanced data with different numbers of data for each cancer type, several methods dealing with imbalanced data are applied to semi-supervised CycleGAN. Results The experimental results showed that the classification performance was enhanced when the dataset from Institution S was used as training data and the testing dataset from Institution T was classified after applying domain transformation. In addition, focal loss contributed to improving the mean F1 score the best as a method that addresses the class imbalance. Conclusion The proposed method achieved the domain transformation of thyroid tissue images between two domains, where it retained the important features related to the classes across domains and showed the best F1 score with significant differences compared with other methods. In addition, the proposed method was further enhanced by addressing the class imbalance of the dataset.</abstract><cop>Cham</cop><pub>Springer International Publishing</pub><pmid>38238492</pmid><doi>10.1007/s11548-024-03061-x</doi><tpages>11</tpages><orcidid>https://orcid.org/0000-0002-8759-8337</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 1861-6429
ispartof International journal for computer assisted radiology and surgery, 2024-11, Vol.19 (11), p.2153-2163
issn 1861-6429
1861-6429
language eng
recordid cdi_proquest_miscellaneous_2929058934
source MEDLINE; SpringerLink Journals - AutoHoldings
subjects Computer Imaging
Computer Science
Deep Learning
Health Informatics
Humans
Image Interpretation, Computer-Assisted - methods
Imaging
Medicine
Medicine & Public Health
Original Article
Pattern Recognition and Graphics
Radiology
Surgery
Thyroid Gland - diagnostic imaging
Thyroid Neoplasms - diagnostic imaging
Thyroid Neoplasms - pathology
Vision
title Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-21T08%3A04%3A20IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Domain%20transformation%20using%20semi-supervised%20CycleGAN%20for%20improving%20performance%20of%20classifying%20thyroid%20tissue%20images&rft.jtitle=International%20journal%20for%20computer%20assisted%20radiology%20and%20surgery&rft.au=Ichiuji,%20Yoshihito&rft.date=2024-11-01&rft.volume=19&rft.issue=11&rft.spage=2153&rft.epage=2163&rft.pages=2153-2163&rft.issn=1861-6429&rft.eissn=1861-6429&rft_id=info:doi/10.1007/s11548-024-03061-x&rft_dat=%3Cproquest_cross%3E2929058934%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2929058934&rft_id=info:pmid/38238492&rfr_iscdi=true