{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T07:10:06Z","timestamp":1778742606358,"version":"3.51.4"},"reference-count":47,"publisher":"IOP Publishing","issue":"9","license":[{"start":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T00:00:00Z","timestamp":1778716800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T00:00:00Z","timestamp":1778716800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"},{"start":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T00:00:00Z","timestamp":1778716800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/publishingsupport.iopscience.iop.org\/questions\/alternative-author-rights-policies\/"},{"start":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T00:00:00Z","timestamp":1778716800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/publishingsupport.iopscience.iop.org\/questions\/alternative-author-rights-policies\/"}],"funder":[{"DOI":"10.13039\/501100003032","name":"Association Nationale de la Recherche et de la Technologie","doi-asserted-by":"crossref","award":["2024\/0145"],"award-info":[{"award-number":["2024\/0145"]}],"id":[{"id":"10.13039\/501100003032","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Phys. Med. Biol."],"published-print":{"date-parts":[[2026,5,14]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>\n                    <jats:italic>Objective.<\/jats:italic>\n                    Inter-reader variability remains a major challenge in breast imaging interpretation, particularly for ordinal classification tasks such as breast density and background parenchymal enhancement (BPE). These visual assessments are prone to inconsistencies due to subjective interpretation, limiting the reliability of AI-based models trained on aggregated or noisy labels. This work aims to account for this variability rather than collapse it.\n                    <jats:italic>Approach.<\/jats:italic>\n                    We propose ReaderAdaptNet, a reader-adaptive network that explicitly models inter-reader variability through reader-specific embeddings. A novel two-stage deep learning framework underpins the training process of the model. The first stage jointly learns discriminative image features and low-dimensional embeddings representing individual annotation styles, enabling personalized classification. The second stage allows for embedding calibration using a small set of labeled examples, enabling rapid adaptation to new readers or consensus definitions without retraining the full model. The method is designed for multi-reader datasets and is evaluated on two breast classification tasks: breast density classification from full-field digital mammography and BPE classification from contrast-enhanced mammography.\n                    <jats:italic>Main results.<\/jats:italic>\n                    Results demonstrate that reader embeddings improve both individual and consensus-level performance, and calibrated embeddings enable flexible, low-cost personalization. Reader-specific embeddings of dimension 32 improved the mean classification accuracy across readers, increasing from 76.4% to 84.4% for breast density, and from 65.1% to 72.1% for BPE, compared to the proposed baseline method.\n                    <jats:italic>Significance.<\/jats:italic>\n                    By disentangling stable image features from reader-specific decision tendencies, ReaderAdaptNet provides a parameter-efficient, interpretable route to\n                    <jats:italic>personalization<\/jats:italic>\n                    (alignment to a given reader) or\n                    <jats:italic>unification<\/jats:italic>\n                    (alignment to an institutional standard) under real-world variability.\n                  <\/jats:p>","DOI":"10.1088\/1361-6560\/ae6227","type":"journal-article","created":{"date-parts":[[2026,4,20]],"date-time":"2026-04-20T22:55:46Z","timestamp":1776725746000},"page":"095032","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["ReaderAdaptNet: modeling reader variability in breast imaging with reader-specific embeddings"],"prefix":"10.1088","volume":"71","author":[{"ORCID":"https:\/\/orcid.org\/0009-0000-2062-058X","authenticated-orcid":true,"given":"Elodie","family":"Ripaud","sequence":"first","affiliation":[{"name":"GE HealthCare"},{"name":"Universit\u00e9 Paris-Saclay"},{"name":"Sorbonne Universit\u00e9"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6414-6039","authenticated-orcid":true,"given":"Cl\u00e9ment","family":"Jailin","sequence":"additional","affiliation":[{"name":"Universit\u00e9 Paris-Saclay"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pablo","family":"Milioni de Carvalho","sequence":"additional","affiliation":[{"name":"GE HealthCare"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-0104-0645","authenticated-orcid":true,"given":"Laurence","family":"Vancamberg","sequence":"additional","affiliation":[{"name":"GE HealthCare"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6984-1532","authenticated-orcid":true,"given":"Isabelle","family":"Bloch","sequence":"additional","affiliation":[{"name":"Sorbonne Universit\u00e9"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"266","published-online":{"date-parts":[[2026,5,14]]},"reference":[{"key":"pmbae6227bib1","doi-asserted-by":"publisher","first-page":"53","DOI":"10.1016\/j.jmir.2018.11.002","type":"journal-article","article-title":"Variability of breast density classification between US and UK radiologists","volume":"50","author":"Alomaim","year":"2019","journal-title":"J. Med. Imaging Radiat. Sci."},{"key":"pmbae6227bib2","doi-asserted-by":"publisher","first-page":"176","DOI":"10.1093\/jbi\/wbaa115","type":"journal-article","article-title":"Training radiologists to interpret contrast-enhanced mammography: toward a standardized lexicon","volume":"3","author":"Berg","year":"2021","journal-title":"J. Breast Imaging"},{"key":"pmbae6227bib3","doi-asserted-by":"publisher","first-page":"1","DOI":"10.48550\/arXiv.1901.02322","type":"conference-proceedings","article-title":"Fusion strategies for learning user embeddings with neural networks","author":"Blandfort","year":"2019"},{"key":"pmbae6227bib4","doi-asserted-by":"publisher","DOI":"10.1097\/MD.0000000000021243","type":"journal-article","article-title":"Fully automatic classification of breast MRI background parenchymal enhancement using a transfer learning approach","volume":"99","author":"Borkowski","year":"2020","journal-title":"Medicine"},{"key":"pmbae6227bib5","doi-asserted-by":"publisher","first-page":"1178","DOI":"10.2214\/AJR.17.18086","type":"journal-article","article-title":"Revisiting nonmass enhancement in breast MRI: analysis of outcomes and follow-up using the updated BI-RADS atlas","volume":"209","author":"Chikarmane","year":"2017","journal-title":"Am. J. Roentgenol."},{"key":"pmbae6227bib6","doi-asserted-by":"publisher","first-page":"114","DOI":"10.1007\/978-3-642-13666-5_16","type":"conference-proceedings","article-title":"A comparative study of the inter-reader variability of breast percent density estimation in digital mammography: potential effect of reader\u2019s training and clinical experience","author":"Conant","year":"2010"},{"key":"pmbae6227bib7","article-title":"ACR BI-RADS Atlas: breast imaging reporting and data system; mammography, ultrasound, magnetic resonance imaging, follow-up and outcome monitoring, data dictionary","author":"D\u2019Orsi","year":"2013","type":"other"},{"key":"pmbae6227bib8","doi-asserted-by":"publisher","first-page":"248","DOI":"10.1109\/CVPR.2009.5206848","type":"conference-proceedings","article-title":"Imagenet: a large-scale hierarchical image database","author":"Deng","year":"2009"},{"key":"pmbae6227bib9","doi-asserted-by":"publisher","first-page":"36","DOI":"10.9734\/ajpas\/2022\/v20i3425","type":"journal-article","article-title":"A comparison of the Pearson, Spearman rank and Kendall tau correlation coefficients using quantitative variables","volume":"20","author":"El-Hashash","year":"2022","journal-title":"Asian J. Probab. Stat."},{"key":"pmbae6227bib10","doi-asserted-by":"publisher","first-page":"1068","DOI":"10.1002\/jmri.28111","type":"journal-article","article-title":"Breast MRI background parenchymal enhancement categorization using deep learning: outperforming the radiologist","volume":"56","author":"Eskreis-Winkler","year":"2022","journal-title":"J. Magn. Resonan. Imaging"},{"key":"pmbae6227bib11","doi-asserted-by":"publisher","first-page":"90","DOI":"10.1186\/s13244-022-01211-w","type":"journal-article","article-title":"Artifact reduction in contrast-enhanced mammography","volume":"13","author":"Gennaro","year":"2022","journal-title":"Insights Imaging"},{"key":"pmbae6227bib12","doi-asserted-by":"publisher","first-page":"1120","DOI":"10.2214\/AJR.14.13047","type":"journal-article","article-title":"Interobserver variability between breast imagers using the 5th edn of the BI-RADS MRI lexicon","volume":"204","author":"Grimm","year":"2015","journal-title":"Am. J. Roentgenol."},{"key":"pmbae6227bib13","doi-asserted-by":"publisher","first-page":"141","DOI":"10.1007\/s10278-018-0114-7","type":"journal-article","article-title":"Fully automated convolutional neural network method for quantification of breast MRI fibroglandular tissue and background parenchymal enhancement","volume":"32","author":"Ha","year":"2019","journal-title":"J. Digit. Imaging"},{"key":"pmbae6227bib14","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/ISBI56570.2024.10635360","type":"conference-proceedings","article-title":"Mammographic breast density classification by integration of deep dictionaries and multi-model sparse approximations","author":"Harris","year":"2024"},{"key":"pmbae6227bib15","first-page":"770","type":"conference-proceedings","article-title":"Deep residual learning for image recognition","author":"He","year":"2016"},{"key":"pmbae6227bib16","doi-asserted-by":"publisher","first-page":"4806","DOI":"10.1109\/ACCESS.2019.2962617","type":"journal-article","article-title":"The real-world-weight cross-entropy loss function: modeling the costs of mislabeling","volume":"8","author":"Ho","year":"2019","journal-title":"IEEE Access"},{"key":"pmbae6227bib17","doi-asserted-by":"publisher","first-page":"49","DOI":"10.1016\/j.breast.2016.06.020","type":"journal-article","article-title":"Consistency of breast density categories in serial screening mammograms: A comparison between automated and human assessment","volume":"29","author":"Holland","year":"2016","journal-title":"The Breast"},{"key":"pmbae6227bib18","doi-asserted-by":"publisher","first-page":"1904","DOI":"10.1145\/3442381.3449933","type":"conference-proceedings","article-title":"Online label aggregation: a variational Bayesian approach","author":"Hong","year":"2021"},{"key":"pmbae6227bib19","doi-asserted-by":"publisher","first-page":"1366","DOI":"10.2214\/AJR.16.16561","type":"journal-article","article-title":"Effects of changes in BI-RADS density assessment guidelines (fourth versus 5th edn) on breast density assessment: Intra- and interreader agreements and density distribution","volume":"207","author":"Irshad","year":"2016","journal-title":"Am. J. Roentgenol."},{"key":"pmbae6227bib20","doi-asserted-by":"publisher","first-page":"340","DOI":"10.2214\/AJR.19.22353","type":"journal-article","article-title":"Factors associated with background parenchymal enhancement on contrast-enhanced mammography","volume":"216","author":"Karimi","year":"2021","journal-title":"Am. J. Roentgenol."},{"key":"pmbae6227bib21","doi-asserted-by":"publisher","first-page":"560","DOI":"10.1016\/j.acra.2013.01.003","type":"journal-article","article-title":"Reader variability in breast density estimation from full-field digital mammograms: the effect of image postprocessing on relative and absolute measures","volume":"20","author":"Keller","year":"2013","journal-title":"Acad. Radiol."},{"key":"pmbae6227bib22","doi-asserted-by":"publisher","first-page":"165","DOI":"10.1148\/radiol.2251011667","type":"journal-article","article-title":"Comparison of the performance of screening mammography, physical examination and breast US and evaluation of factors that influence them: an analysis of 27,825 patient evaluations","volume":"225","author":"Kolb","year":"2002","journal-title":"Radiology"},{"key":"pmbae6227bib23","author":"Lee","year":"2022","type":"book"},{"key":"pmbae6227bib24","doi-asserted-by":"publisher","first-page":"52","DOI":"10.1148\/radiol.2018180694","type":"journal-article","article-title":"Mammographic breast density assessment using deep learning: clinical implementation","volume":"290","author":"Lehman","year":"2019","journal-title":"Radiology"},{"key":"pmbae6227bib25","doi-asserted-by":"publisher","first-page":"5879","DOI":"10.1109\/TNNLS.2024.3396463","type":"journal-article","article-title":"A survey on deep active learning: Recent advances and new frontiers","volume":"36","author":"Li","year":"2024","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"pmbae6227bib26","doi-asserted-by":"publisher","first-page":"1045","DOI":"10.1109\/TMM.2021.3137752","type":"journal-article","article-title":"Trustable co-label learning from multiple noisy annotators","volume":"25","author":"Li","year":"2021","journal-title":"IEEE Trans. Multimedia"},{"key":"pmbae6227bib27","doi-asserted-by":"publisher","first-page":"6922","DOI":"10.1109\/ACCESS.2023.3237990","type":"journal-article","article-title":"Deep Gaussian processes for classification with multiple noisy annotators. Application to breast cancer tissue classification","volume":"11","author":"L\u00f3pez-P\u00e9rez","year":"2023","journal-title":"IEEE Access"},{"key":"pmbae6227bib28","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.210199","type":"journal-article","article-title":"Development and validation of an AI-driven mammographic breast density classification tool based on radiologist consensus","volume":"4","author":"Magni","year":"2022","journal-title":"Radiology"},{"key":"pmbae6227bib29","doi-asserted-by":"publisher","first-page":"209","DOI":"10.2214\/AJR.13.10952","type":"journal-article","article-title":"Inter-and intrareader agreement for categorization of background parenchymal enhancement at baseline and after training","volume":"203","author":"Melsaether","year":"2014","journal-title":"Am. J. Roentgenol."},{"key":"pmbae6227bib30","doi-asserted-by":"publisher","first-page":"818","DOI":"10.1002\/jmri.27429","type":"journal-article","article-title":"Fully automatic assessment of background parenchymal enhancement on breast MRI using machine-learning models","volume":"53","author":"Nam","year":"2021","journal-title":"J. Magn. Resonan. Imaging"},{"key":"pmbae6227bib31","doi-asserted-by":"publisher","first-page":"1623","DOI":"10.1109\/ISBI48211.2021.9433965","type":"conference-proceedings","article-title":"Confidence-guided learning for breast density classification","author":"Park","year":"2021"},{"key":"pmbae6227bib32","doi-asserted-by":"publisher","first-page":"671","DOI":"10.1016\/j.jacr.2023.03.010","type":"journal-article","article-title":"International interobserver variability of breast density assessment","volume":"20","author":"Portnow","year":"2023","journal-title":"J. Am. Coll. Radiol."},{"key":"pmbae6227bib33","doi-asserted-by":"publisher","first-page":"21","DOI":"10.1016\/j.clinimag.2021.11.034","type":"journal-article","article-title":"Persistent inter-observer variability of breast density assessment using BI-RADS\u00ae 5th edn guidelines","volume":"83","author":"Portnow","year":"2022","journal-title":"Clin. Imaging"},{"key":"pmbae6227bib34","doi-asserted-by":"publisher","first-page":"1465","DOI":"10.1259\/bjr\/21256379","type":"journal-article","article-title":"Inter-and intraradiologist variability in the BI-RADS assessment and breast density categories for screening mammograms","volume":"85","author":"Redondo","year":"2012","journal-title":"Br. J. Radiol."},{"key":"pmbae6227bib35","doi-asserted-by":"publisher","first-page":"130","DOI":"10.1007\/978-3-032-04981-0_13","type":"conference-proceedings","article-title":"Deep learning framework for managing inter-reader variability in background parenchymal enhancement classification for contrast-enhanced mammography","author":"Ripaud","year":"2025"},{"key":"pmbae6227bib36","doi-asserted-by":"publisher","first-page":"471","DOI":"10.1117\/12.3025536","type":"conference-proceedings","article-title":"Deep-learning-based background parenchymal enhancement quantification in contrast enhanced mammography: an application to neoadjuvant chemotherapy","volume":"vol 13174","author":"Ripaud","year":"2024a"},{"key":"pmbae6227bib37","doi-asserted-by":"publisher","DOI":"10.1088\/1361-6560\/ad42ff","type":"journal-article","article-title":"Deep-learning model for background parenchymal enhancement classification in contrast-enhanced mammography","volume":"69","author":"Ripaud","year":"2024b","journal-title":"Phys. Med. Biol."},{"key":"pmbae6227bib38","doi-asserted-by":"publisher","first-page":"839","DOI":"10.3390\/curroncol30010064","type":"journal-article","article-title":"Machine learning approaches with textural features to calculate breast density on mammography","volume":"30","author":"Sansone","year":"2023","journal-title":"Curr. Oncol."},{"key":"pmbae6227bib39","doi-asserted-by":"publisher","first-page":"1234","DOI":"10.1016\/j.acra.2019.10.034","type":"journal-article","article-title":"Background parenchymal enhancement at contrast-enhanced spectral mammography (CESM) as a breast cancer risk factor","volume":"27","author":"Sorin","year":"2020","journal-title":"Acad. Radiol."},{"key":"pmbae6227bib40","doi-asserted-by":"publisher","first-page":"4666","DOI":"10.1002\/sim.5848","type":"journal-article","article-title":"Correcting for rater bias in scores on a continuous scale, with application to breast density","volume":"32","author":"Sperrin","year":"2013","journal-title":"Stat. Med."},{"key":"pmbae6227bib41","doi-asserted-by":"publisher","DOI":"10.1088\/2057-1976\/accaea","type":"journal-article","article-title":"The effect of variable labels on deep learning models trained to predict breast density","volume":"9","author":"Squires","year":"2023","journal-title":"Biomed. Phys. Eng. Express"},{"key":"pmbae6227bib42","doi-asserted-by":"publisher","first-page":"682","DOI":"10.3390\/math11030682","type":"journal-article","article-title":"Recent advances in stochastic gradient descent in deep learning","volume":"11","author":"Tian","year":"2023","journal-title":"Mathematics"},{"key":"pmbae6227bib43","doi-asserted-by":"publisher","first-page":"569","DOI":"10.1016\/j.cmpb.2013.10.004","type":"journal-article","article-title":"Breast density classification to reduce false positives in CADe systems","volume":"113","author":"V\u00e1llez","year":"2014","journal-title":"Comput. Methods Programs Biomed."},{"key":"pmbae6227bib44","first-page":"360","type":"journal-article","article-title":"Understanding interobserver agreement: the kappa statistic","volume":"37","author":"Viera","year":"2005","journal-title":"Family Med."},{"key":"pmbae6227bib45","doi-asserted-by":"publisher","first-page":"548","DOI":"10.1016\/j.mayocp.2013.12.014","type":"conference-proceedings","article-title":"Breast density and breast cancer risk: a practical review","volume":"vol 89","author":"Wang","year":"2014"},{"key":"pmbae6227bib46","doi-asserted-by":"publisher","first-page":"2694","DOI":"10.3390\/diagnostics13162694","type":"journal-article","article-title":"Mammographic breast density model using semi-supervised learning reduces inter-\/intra-reader variability","volume":"13","author":"Watanabe","year":"2023","journal-title":"Diagnostics"},{"key":"pmbae6227bib47","doi-asserted-by":"publisher","first-page":"291","DOI":"10.1007\/s10994-013-5412-1","type":"journal-article","article-title":"Learning from multiple annotators with varying expertise","volume":"95","author":"Yan","year":"2014","journal-title":"Mach. Learn."}],"container-title":["Physics in Medicine &amp; Biology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T06:38:10Z","timestamp":1778740690000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae6227"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,14]]},"references-count":47,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2026,5,14]]},"published-print":{"date-parts":[[2026,5,14]]}},"URL":"https:\/\/doi.org\/10.1088\/1361-6560\/ae6227","relation":{},"ISSN":["0031-9155","1361-6560"],"issn-type":[{"value":"0031-9155","type":"print"},{"value":"1361-6560","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,14]]},"assertion":[{"value":"ReaderAdaptNet: modeling reader variability in breast imaging with reader-specific embeddings","name":"article_title","label":"Article Title"},{"value":"Physics in Medicine & Biology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2026 The Author(s). Published on behalf of Institute of Physics and Engineering in Medicine by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2025-11-10","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2026-04-20","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2026-05-14","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}