A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification

<!DOCTYPE html> <html lang="en" xml:lang="en"> <head> <meta charset="utf-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <title> A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification | U.Porto Journal of Engineering </title> <link rel="icon" href="https://journalengineering.fe.up.pt/public/journals/3/favicon_en_US.png"> <meta name="generator" content="Open Journal Systems 3.4.0.3"> <link rel="schema.DC" href="http://purl.org/dc/elements/1.1/" /> <meta name="DC.Creator.PersonalName" content="Joana Rocha"/> <meta name="DC.Creator.PersonalName" content="Ana Maria Mendonça"/> <meta name="DC.Creator.PersonalName" content="Aurélio Campilho"/> <meta name="DC.Date.created" scheme="ISO8601" content="2021-11-26"/> <meta name="DC.Date.dateSubmitted" scheme="ISO8601" content="2020-11-27"/> <meta name="DC.Date.issued" scheme="ISO8601" content="2021-11-26"/> <meta name="DC.Date.modified" scheme="ISO8601" content="2021-11-26"/> <meta name="DC.Description" xml:lang="en" content="Backed by more powerful computational resources and optimized training routines, deep learning models have proven unprecedented performance and several benefits to extract information from chest X-ray data. This is one of the most common imaging exams, whose increasing demand is reflected in the aggravated radiologists’ workload. Consequently, healthcare would benefit from computer-aided diagnosis systems to prioritize certain exams and further identify possible pathologies. Pioneering work in chest X-ray analysis has focused on the identification of specific diseases, but to the best of the authors' knowledge no paper has specifically reviewed relevant work on abnormality detection and multi-label thoracic pathology classification. This paper focuses on those issues, selecting the leading chest X-ray based deep learning strategies for comparison. In addition, the paper discloses the current annotated public chest X-ray databases, covering the common thorax diseases."/> <meta name="DC.Format" scheme="IMT" content="application/pdf"/> <meta name="DC.Identifier" content="2183-6493_007-004_0002"/> <meta name="DC.Identifier.pageNumber" content="16-32"/> <meta name="DC.Identifier.DOI" content="10.24840/2183-6493_007.004_0002"/> <meta name="DC.Identifier.URI" content="https://journalengineering.fe.up.pt/index.php/upjeng/article/view/2183-6493_007-004_0002"/> <meta name="DC.Language" scheme="ISO639-1" content="en"/> <meta name="DC.Rights" content="Copyright (c) 2021 Joana Rocha, Ana Maria Mendonça, Aurélio Campilho"/> <meta name="DC.Rights" content="https://creativecommons.org/licenses/by/4.0"/> <meta name="DC.Source" content="U.Porto Journal of Engineering"/> <meta name="DC.Source.ISSN" content="2183-6493"/> <meta name="DC.Source.Issue" content="4"/> <meta name="DC.Source.Volume" content="7"/> <meta name="DC.Source.URI" content="https://journalengineering.fe.up.pt/index.php/upjeng"/> <meta name="DC.Subject" xml:lang="en" content="Computer-aided Diagnosis"/> <meta name="DC.Subject" xml:lang="en" content="Deep Neural Network"/> <meta name="DC.Subject" xml:lang="en" content="Medical Imaging"/> <meta name="DC.Subject" xml:lang="en" content="Radiology"/> <meta name="DC.Subject" xml:lang="en" content="Thorax"/> <meta name="DC.Title" content="A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification"/> <meta name="DC.Type" content="Text.Serial.Journal"/> <meta name="DC.Type.articleType" content="PEC202021_01"/> <meta name="gs_meta_revision" content="1.1"/> <meta name="citation_journal_title" content="U.Porto Journal of Engineering"/> <meta name="citation_journal_abbrev" content="UPjeng"/> <meta name="citation_issn" content="2183-6493"/> <meta name="citation_author" content="Joana Rocha"/> <meta name="citation_author_institution" content="INESC TEC; University of Porto"/> <meta name="citation_author" content="Ana Maria Mendonça"/> <meta name="citation_author_institution" content="INESC TEC; University of Porto"/> <meta name="citation_author" content="Aurélio Campilho"/> <meta name="citation_author_institution" content="INESC TEC; University of Porto"/> <meta name="citation_title" content="A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification"/> <meta name="citation_language" content="en"/> <meta name="citation_date" content="2021/11/26"/> <meta name="citation_volume" content="7"/> <meta name="citation_issue" content="4"/> <meta name="citation_firstpage" content="16"/> <meta name="citation_lastpage" content="32"/> <meta name="citation_doi" content="10.24840/2183-6493_007.004_0002"/> <meta name="citation_abstract_html_url" content="https://journalengineering.fe.up.pt/index.php/upjeng/article/view/2183-6493_007-004_0002"/> <meta name="citation_abstract" xml:lang="en" content="Backed by more powerful computational resources and optimized training routines, deep learning models have proven unprecedented performance and several benefits to extract information from chest X-ray data. This is one of the most common imaging exams, whose increasing demand is reflected in the aggravated radiologists’ workload. Consequently, healthcare would benefit from computer-aided diagnosis systems to prioritize certain exams and further identify possible pathologies. Pioneering work in chest X-ray analysis has focused on the identification of specific diseases, but to the best of the authors' knowledge no paper has specifically reviewed relevant work on abnormality detection and multi-label thoracic pathology classification. This paper focuses on those issues, selecting the leading chest X-ray based deep learning strategies for comparison. In addition, the paper discloses the current annotated public chest X-ray databases, covering the common thorax diseases."/> <meta name="citation_keywords" xml:lang="en" content="Computer-aided Diagnosis"/> <meta name="citation_keywords" xml:lang="en" content="Deep Neural Network"/> <meta name="citation_keywords" xml:lang="en" content="Medical Imaging"/> <meta name="citation_keywords" xml:lang="en" content="Radiology"/> <meta name="citation_keywords" xml:lang="en" content="Thorax"/> <meta name="citation_pdf_url" content="https://journalengineering.fe.up.pt/index.php/upjeng/article/download/2183-6493_007-004_0002/565"/> <link rel="alternate" type="application/atom+xml" href="https://journalengineering.fe.up.pt/index.php/upjeng/gateway/plugin/APP%5Cplugins%5Cgeneric%5CwebFeed%5CWebFeedGatewayPlugin/atom"> <link rel="alternate" type="application/rdf+xml" href="https://journalengineering.fe.up.pt/index.php/upjeng/gateway/plugin/APP%5Cplugins%5Cgeneric%5CwebFeed%5CWebFeedGatewayPlugin/rss"> <link rel="alternate" type="application/rss+xml" href="https://journalengineering.fe.up.pt/index.php/upjeng/gateway/plugin/APP%5Cplugins%5Cgeneric%5CwebFeed%5CWebFeedGatewayPlugin/rss2"> <link rel="stylesheet" href="https://journalengineering.fe.up.pt/index.php/upjeng/$$$call$$$/page/page/css?name=bootstrap" type="text/css" /> </head> <body class="pkp_page_article pkp_op_view has_site_logo"> <div class="pkp_structure_page"> <nav id="accessibility-nav" class="sr-only" role="navigation" aria-label="Quick jump to page content"> <ul> <li><a href="#main-navigation">Main Navigation</a></li> <li><a href="#main-content">Main Content</a></li> <li><a href="#sidebar">Sidebar</a></li> </ul> </nav> <header class="navbar navbar-default" id="headerNavigationContainer" role="banner"> <div class="container-fluid"> <div class="row"> <nav aria-label="User Navigation"> <ul id="navigationUser" class="nav nav-pills tab-list pull-right"> <li class=" menu-item-1"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/user/register"> Register </a> </li> <li class=" menu-item-2"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/login"> Login </a> </li> </ul> </nav> </div> </div> <div class="container-fluid"> <div class="navbar-header"> <button type="button" class="navbar-toggle collapsed" data-toggle="collapse" data-target="#nav-menu" aria-expanded="false" aria-controls="nav-menu"> <span class="sr-only">Toggle navigation</span> <span class="icon-bar"></span> <span class="icon-bar"></span> <span class="icon-bar"></span> </button> <div class="site-name"> <a href=" https://journalengineering.fe.up.pt/index.php/upjeng/index " class="navbar-brand navbar-brand-logo"> <img src="https://journalengineering.fe.up.pt/public/journals/3/pageHeaderLogoImage_en_US.png" alt="U. Porto Journal of Engineering"> </a> </div> </div> <nav id="nav-menu" class="navbar-collapse collapse" aria-label="Site Navigation"> <ul id="main-navigation" class="nav navbar-nav"> <li class=" menu-item-8"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/issue/current"> Current </a> </li> <li class=" menu-item-9"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/issue/archive"> Archives </a> </li> <li class=" menu-item-11"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/about"> About </a> </li> <li class=" menu-item-14"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/about/editorialTeam"> Editorial Team </a> </li> <li class=" menu-item-13"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/about/submissions"> Submissions </a> </li> <li class=" menu-item-53"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/openaccess"> Open Access Policy </a> </li> <li class=" menu-item-54"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/peerreview"> Peer Review </a> </li> <li class=" menu-item-55"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/ethics"> Publication Ethics </a> </li> <li class=" menu-item-15"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/about/contact"> Contact </a> </li> </ul> <div class="pull-md-right"> <form class="navbar-form navbar-left" role="search" method="post" action="https://journalengineering.fe.up.pt/index.php/upjeng/search/search"> <div class="form-group"> <input class="form-control" name="query" value="" type="search" aria-label="Search Query" placeholder=""> </div> <button type="submit" class="btn btn-default">Search</button> </form> </div> </nav> </div> </header> <div class="pkp_structure_content container"> <main class="pkp_structure_main col-xs-12 col-sm-10 col-md-12" role="main"> <div class="page page_article"> <nav class="cmp_breadcrumbs" role="navigation" aria-label="You are here:"> <ol class="breadcrumb"> <li> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/index"> Home </a> </li> <li> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/issue/archive"> Archives </a> </li> <li> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/issue/view/2183-6493_007-004"> Vol. 7 No. 4 (2021) </a> </li> <li class="active"> PEC202021_01 </li> </ol> </nav> <article class="article-details"> <header> <h1 class="page-header"> A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification </h1> </header> <div class="row"> <section class="article-sidebar col-md-4"> <h2 class="sr-only">Article Sidebar</h2> <div class="cover-image"> <img class="img-responsive" src="https://journalengineering.fe.up.pt/public/journals/3/submission_774_623_coverImage_en_US.jpg" alt="A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification" > </div> <div class="download"> <a class="galley-link btn btn-primary pdf" role="button" href="https://journalengineering.fe.up.pt/index.php/upjeng/article/view/2183-6493_007-004_0002/565"> PDF </a> </div> <div class="list-group"> <div class="list-group-item date-published"> <strong>Published:</strong> Nov 26, 2021 </div> <div class="list-group-item doi"> <strong>DOI:</strong> <a href="https://doi.org/10.24840/2183-6493_007.004_0002"> https://doi.org/10.24840/2183-6493_007.004_0002 </a> </div> <div class="list-group-item issue"> <strong> Issue: </strong> <span class=""> <a class="title" href="https://journalengineering.fe.up.pt/index.php/upjeng/issue/view/2183-6493_007-004"> Vol. 7 No. 4 (2021) </a> </span> </div> <div class="list-group-item keywords"> <strong> Keywords:</strong> <div class=""> <span class="value"> Computer-aided Diagnosis, Deep Neural Network, Medical Imaging, Radiology, Thorax </span> </div> </div> </div> </section> <div class="col-md-8"> <section class="article-main"> <h2 class="sr-only">Main Article Content</h2> <div class="authors"> <div class="author"> <strong>Joana Rocha</strong> <div class="article-author-affilitation"> INESC TEC; University of Porto </div> <div class="orcid"> <a href="https://orcid.org/0000-0002-4856-138X" target="_blank"> https://orcid.org/0000-0002-4856-138X </a> </div> </div> <div class="author"> <strong>Ana Maria Mendonça</strong> <div class="article-author-affilitation"> INESC TEC; University of Porto </div> <div class="orcid"> <a href="https://orcid.org/0000-0002-4319-738X" target="_blank"> https://orcid.org/0000-0002-4319-738X </a> </div> </div> <div class="author"> <strong>Aurélio Campilho</strong> <div class="article-author-affilitation"> INESC TEC; University of Porto </div> <div class="orcid"> <a href="https://orcid.org/0000-0002-5317-6275" target="_blank"> https://orcid.org/0000-0002-5317-6275 </a> </div> </div> </div> <div class="article-summary" id="summary"> <h2>Abstract</h2> <div class="article-abstract"> <p>Backed by more powerful computational resources and optimized training routines, deep learning models have proven unprecedented performance and several benefits to extract information from chest X-ray data. This is one of the most common imaging exams, whose increasing demand is reflected in the aggravated radiologists’ workload. Consequently, healthcare would benefit from computer-aided diagnosis systems to prioritize certain exams and further identify possible pathologies. Pioneering work in chest X-ray analysis has focused on the identification of specific diseases, but to the best of the authors' knowledge no paper has specifically reviewed relevant work on abnormality detection and multi-label thoracic pathology classification. This paper focuses on those issues, selecting the leading chest X-ray based deep learning strategies for comparison. In addition, the paper discloses the current annotated public chest X-ray databases, covering the common thorax diseases.</p> </div> </div> <section class="item downloads_chart"> <h2 class="label"> Downloads </h2> <div class="value"> <canvas class="usageStatsGraph" data-object-type="Submission" data-object-id="774"></canvas> <div class="usageStatsUnavailable" data-object-type="Submission" data-object-id="774"> Download data is not yet available. </div> </div> </section> </section> <section class="article-more-details"> <h2 class="sr-only">Article Details</h2>   <div class="panel panel-default copyright"> <div class="panel-body"> <a rel="license" href="https://creativecommons.org/licenses/by/4.0/"><img alt="Creative Commons License" src="//i.creativecommons.org/l/by/4.0/88x31.png" /></a><p>This work is licensed under a <a rel="license" href="https://creativecommons.org/licenses/by/4.0/">Creative Commons Attribution 4.0 International License</a>.</p> <p>Authors who publish with this journal agree to the following terms:</p> <ol type="a"> <li>Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">Creative Commons Attribution License</a> that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.</li> <li>Authors grant the journal the rights to provide the article in all forms and media so the article can be used on the latest technology even after publication and ensure its long-term preservation.</li> <li>Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.</li> <li>Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See <a href="http://opcit.eprints.org/oacitation-biblio.html" target="_blank">The Effect of Open Access</a>).</li> </ol> </div> </div> <div class="panel panel-default author-bios"> <div class="panel-heading"> Author Biographies </div> <div class="panel-body"> <div class="media biography"> <div class="media-body"> <h3 class="media-heading biography-author"> Joana Rocha, <span class="affiliation">INESC TEC; University of Porto</span> </h3> <p>INESC TEC-Institute for Systems and Computer Engineering</p> <p>Faculty of Engineering</p> <p>University of Porto</p> <p>Rua Dr. Roberto Frias</p> <p>4200-465 PORTO</p> <p>Portugal</p> </div> </div> <div class="media biography"> <div class="media-body"> <h3 class="media-heading biography-author"> Ana Maria Mendonça, <span class="affiliation">INESC TEC; University of Porto</span> </h3> <p>INESC TEC-Institute for Systems and Computer Engineering</p> <p>Faculty of Engineering</p> <p>University of Porto</p> <p>Rua Dr. Roberto Frias</p> <p>4200-465 PORTO</p> <p>Portugal</p> </div> </div> <div class="media biography"> <div class="media-body"> <h3 class="media-heading biography-author"> Aurélio Campilho, <span class="affiliation">INESC TEC; University of Porto</span> </h3> <p>INESC TEC-Institute for Systems and Computer Engineering</p> <p>Faculty of Engineering</p> <p>University of Porto</p> <p>Rua Dr. Roberto Frias</p> <p>4200-465 PORTO</p> <p>Portugal</p> </div> </div> </div> </div> </section> </div> </div> </article> </div> </main> </div> <footer class="footer" role="contentinfo"> <div class="container"> <div class="row"> <div class="col-md-10"> <p><small><a href="https://www.fe.up.pt" target="_blank" rel="noopener">Faculdade de Engenharia da Universidade do Porto</a><br /><a href="https://biblioteca.fe.up.pt" target="_blank" rel="noopener">Serviços de Documentação e Informação: Biblioteca</a><br />Rua Dr. Roberto Frias<br />4200-465 PORTO<br /><a title="Biblioteca da FEUP" href="mailto:biblioteca@fe.up.pt">biblioteca@fe.up.pt</a> |<a href="tel:+351220413805"> +351 220413805</a></small></p> <p>ISSN <a title="ROAD information" href="https://portal.issn.org/resource/ISSN/2183-6493" target="_blank" rel="noopener">2183-6493</a><br />DOI <a href="https://doi.org/10.24840/2183-6493">10.24840/2183-6493</a></p> </div> <div class="col-md-2" role="complementary"> <a href="https://journalengineering.fe.up.pt/index.php/upjeng/about/aboutThisPublishingSystem"> <img class="img-responsive" alt="More information about the publishing system, Platform and Workflow by OJS/PKP." src="https://journalengineering.fe.up.pt/templates/images/ojs_brand.png"> </a> </div> </div>  </div> </footer> </div> <script src="https://journalengineering.fe.up.pt/lib/pkp/lib/vendor/components/jquery/jquery.min.js?v=3.4.0.3" type="text/javascript"></script><script src="https://journalengineering.fe.up.pt/lib/pkp/lib/vendor/components/jqueryui/jquery-ui.min.js?v=3.4.0.3" type="text/javascript"></script><script src="https://journalengineering.fe.up.pt/lib/pkp/js/lib/jquery/plugins/jquery.tag-it.js?v=3.4.0.3" type="text/javascript"></script><script src="https://journalengineering.fe.up.pt/plugins/themes/bootstrap3/bootstrap/js/bootstrap.min.js?v=3.4.0.3" type="text/javascript"></script><script src="https://journalengineering.fe.up.pt/plugins/themes/bootstrap3/bootstrap/js/cookie.js?v=3.4.0.3" type="text/javascript"></script><script type="text/javascript">var pkpUsageStats = pkpUsageStats || {};pkpUsageStats.data = pkpUsageStats.data || {};pkpUsageStats.data.Submission = pkpUsageStats.data.Submission || {};pkpUsageStats.data.Submission[774] = {"data":{"2021":{"11":"38","12":"113"},"2022":{"1":"125","2":"92","3":"72","4":"38","5":"63","6":"65","7":"19","8":"14","9":"13","10":"12","11":"13","12":"20"},"2023":{"1":"20","2":"12","3":"19","4":"10","5":"18","6":"8","7":"12","8":"14","9":"9","10":"11","11":"7","12":"9"},"2024":{"1":"1","2":"4","3":"7","4":"8","5":"10","6":"10","7":"11","8":"13","9":"9","10":"7","11":"3"}},"label":"All Downloads","color":"79,181,217","total":929};</script><script src="https://journalengineering.fe.up.pt/lib/pkp/js/lib/Chart.min.js?v=3.4.0.3" type="text/javascript"></script><script type="text/javascript">var pkpUsageStats = pkpUsageStats || {};pkpUsageStats.locale = pkpUsageStats.locale || {};pkpUsageStats.locale.months = ["Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec"];pkpUsageStats.config = pkpUsageStats.config || {};pkpUsageStats.config.chartType = "line";</script><script src="https://journalengineering.fe.up.pt/lib/pkp/js/usage-stats-chart.js?v=3.4.0.3" type="text/javascript"></script><script type="text/javascript"> (function (w, d, s, l, i) { w[l] = w[l] || []; var f = d.getElementsByTagName(s)[0], j = d.createElement(s), dl = l != 'dataLayer' ? '&l=' + l : ''; j.async = true; j.src = 'https://www.googletagmanager.com/gtag/js?id=' + i + dl; f.parentNode.insertBefore(j, f); function gtag(){dataLayer.push(arguments)}; gtag('js', new Date()); gtag('config', i); }) (window, document, 'script', 'dataLayer', 'UA-26693817-4'); </script>  <div id="cookie_directive_container" class="container" style="display: none"> <nav class="navbar navbar-default navbar-fixed-bottom"> <div class="container"> <div class="navbar-inner navbar-content-center" id="cookie_accept"> <a href="#" class="btn btn-default pull-right">Close</a> <p class="text-muted credit"> By using our website you are consenting to our use of cookies in accordance with our <a href="/index.php/index/privacy" target="_blank">cookies policy</a>. </p> </div> </div> </nav> </div> </body> </html>

CINXE.COM

A Review on Deep Learning Methods for Chest X-Ray based Abnormality Detection and Thoracic Pathology Classification | U.Porto Journal of Engineering