{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T13:22:15Z","timestamp":1787750535585,"version":"build-2784847793"},"reference-count":35,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T00:00:00Z","timestamp":1787702400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T00:00:00Z","timestamp":1787702400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Insights Imaging"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Objectives<\/jats:title>\n                    <jats:p>To evaluate an artificial intelligence (AI) algorithm on three different mammography devices and assess screening performance using general and device-specific thresholds.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Materials and methods<\/jats:title>\n                    <jats:p>This retrospective study analyzed 22,673 AI-assessed mammographies of women aged 50\u201369 collected in 2022\u20132023 from three mammography devices across six screening sites within a Swiss Mammography Screening Program. 115 screen-detected and 15 1-year interval breast cancers were included in the analyses. A commercially available AI algorithm assessed each mammography with a numeric case score. Optimal thresholds were determined across all devices and for each device separately using the Youden-Index. Screening performance was assessed using sensitivity, specificity, balanced accuracy, and cancer detection rate (CDR).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      Standard double-reading had a CDR of 5.07 (95% CI: 4.19\u20136.09). Using AI with a general threshold achieved a CDR of 4.81 (95% CI: 3.95\u20135.80), which was enhanced using device-specific thresholds to 4.90 (95% CI: 4.03\u20135.89). Workload for radiologists in consensus conferences was lowered by nearly one-third when using device-specific thresholds compared to a general threshold (\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001). Nonetheless, AI screening performance for one of the mammography devices was poor even with device-specific thresholds.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>AI-based performance varied considerably among the three mammography devices, emphasizing the need for device-specific thresholds or even the withdrawal of AI use on certain devices. Awareness of different case distribution among radiologists should be raised to mitigate the risk of interpretation biases and reduced diagnostic accuracy. Lastly, training datasets must be representative not only of the target population but also of the different screening devices used in practice.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Key Points<\/jats:title>\n                    <jats:p>\n                      <jats:bold>Question<\/jats:bold>\n                      AI performance may vary across mammography devices, raising uncertainty about device-independent thresholds.\n                    <\/jats:p>\n                    <jats:p>\n                      <jats:bold>Findings<\/jats:bold>\n                      AI scores differed considerably across devices, indicating the need for device-specific thresholds, but performance on one device remained poor.\n                    <\/jats:p>\n                    <jats:p>\n                      <jats:bold>Critical relevance<\/jats:bold>\n                      The performance of an AI algorithm in mammography screening depends on adequately defined decision thresholds, and radiologists must be aware that AI scores can vary across mammography devices to avoid interpretation biases and to maintain diagnostic accuracy.\n                    <\/jats:p>\n                  <\/jats:sec>","DOI":"10.1186\/s13244-026-02383-5","type":"journal-article","created":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T12:52:30Z","timestamp":1787748750000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["AI performance varies considerably across mammography devices: a multi-site and multi-vendor retrospective study"],"prefix":"10.1186","volume":"17","author":[{"ORCID":"https:\/\/orcid.org\/0009-0008-9243-3896","authenticated-orcid":false,"given":"Marcel","family":"Blum","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rudolf","family":"Morant","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alena","family":"Eichenberger","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alexander","family":"Geissler","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jonas","family":"Subelack","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Justus","family":"Vogel","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"Ehlig","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,8,26]]},"reference":[{"key":"2383_CR1","doi-asserted-by":"publisher","DOI":"10.1186\/s13058-024-01841-6","volume":"26","author":"D Kuklinski","year":"2024","unstructured":"Kuklinski D, Blum M, Subelack J, Geissler A, Eichenberger A, Morant R (2024) Breast cancer patients enrolled in the Swiss mammography screening program \u201cdonna\u201d demonstrate prolonged survival. Breast Cancer Res 26:84","journal-title":"Breast Cancer Res"},{"key":"2383_CR2","doi-asserted-by":"publisher","DOI":"10.1186\/s12885-021-07917-2","volume":"21","author":"C Herrmann","year":"2021","unstructured":"Herrmann C, Morant R, Walser E, Mousavi M, Thurlimann B (2021) Screening is associated with lower mastectomy rates in eastern Switzerland beyond stage effects. BMC Cancer 21:229","journal-title":"BMC Cancer"},{"key":"2383_CR3","doi-asserted-by":"publisher","first-page":"5974","DOI":"10.1007\/s00330-022-08711-9","volume":"32","author":"MA Martiniussen","year":"2022","unstructured":"Martiniussen MA, Sagstad S, Larsen M et al (2022) Screen-detected and interval breast cancer after concordant and discordant interpretations in a population based screening program using independent double reading. Eur Radiol 32:5974\u20135985","journal-title":"Eur Radiol"},{"key":"2383_CR4","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-025-57409-1","volume":"16","author":"ME Ahsen","year":"2025","unstructured":"Ahsen ME, Ayvaci MUS, Mookerjee R, Stolovitzky G (2025) Economics of AI and human task sharing for decision making in screening mammography. Nat Commun 16:2289","journal-title":"Nat Commun"},{"key":"2383_CR5","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.230529","volume":"6","author":"MT Elhakim","year":"2024","unstructured":"Elhakim MT, Stougaard SW, Graumann O et al (2024) AI-integrated screening to replace double reading of mammograms: a population-wide accuracy and feasibility study. Radiol Artif Intell 6:e230529","journal-title":"Radiol Artif Intell"},{"key":"2383_CR6","doi-asserted-by":"crossref","unstructured":"Dembrower K, Crippa A, Colon E, Eklund M, Strand F, ScreenTrustCAD Trial Consortium (2023) Artificial intelligence for breast cancer detection in screening mammography in Sweden: a prospective, population-based, paired-reader, non-inferiority study. Lancet Digit Health 5:e703\u2013e711","DOI":"10.1016\/S2589-7500(23)00153-X"},{"key":"2383_CR7","doi-asserted-by":"publisher","first-page":"e175","DOI":"10.1016\/S2589-7500(24)00267-X","volume":"7","author":"V Hernstrom","year":"2025","unstructured":"Hernstrom V, Josefsson V, Sartor H et al (2025) Screening performance and characteristics of breast cancer detected in the Mammography Screening with Artificial Intelligence trial (MASAI): a randomised, controlled, parallel-group, non-inferiority, single-blinded, screening accuracy study. Lancet Digit Health 7:e175\u2013e183","journal-title":"Lancet Digit Health"},{"key":"2383_CR8","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.232479","volume":"311","author":"AD Lauritzen","year":"2024","unstructured":"Lauritzen AD, Lillholm M, Lynge E, Nielsen M, Karssemeijer N, Vejborg I (2024) Early indicators of the impact of using AI in mammography screening for breast cancer. Radiology 311:e232479","journal-title":"Radiology"},{"key":"2383_CR9","doi-asserted-by":"publisher","first-page":"917","DOI":"10.1038\/s41591-024-03408-6","volume":"31","author":"N Eisemann","year":"2025","unstructured":"Eisemann N, Bunk S, Mukama T et al (2025) Nationwide real-world implementation of AI for cancer detection in population-based mammography screening. Nat Med 31:917\u2013924","journal-title":"Nat Med"},{"key":"2383_CR10","doi-asserted-by":"publisher","first-page":"e803","DOI":"10.1016\/S2589-7500(24)00173-0","volume":"6","author":"ZV Fisches","year":"2024","unstructured":"Fisches ZV, Ball M, Mukama T et al (2024) Strategies for integrating artificial intelligence into mammography screening programmes: a retrospective simulation analysis. Lancet Digit Health 6:e803\u2013e814","journal-title":"Lancet Digit Health"},{"key":"2383_CR11","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-024-51725-8","volume":"15","author":"HML Frazer","year":"2024","unstructured":"Frazer HML, Pena-Solorzano CA, Kwok CF et al (2024) Comparison of AI-integrated pathways with human-AI interaction in population mammographic screening for breast cancer. Nat Commun 15:7525","journal-title":"Nat Commun"},{"key":"2383_CR12","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.222639","volume":"307","author":"JH Yoon","year":"2023","unstructured":"Yoon JH, Strand F, Baltzer PAT et al (2023) Standalone AI for breast cancer detection at screening digital mammography and digital breast tomosynthesis: a systematic review and meta-analysis. Radiology 307:e222639","journal-title":"Radiology"},{"key":"2383_CR13","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1038\/s41416-021-01333-w","volume":"125","author":"SE Hickman","year":"2021","unstructured":"Hickman SE, Baxter GC, Gilbert FJ (2021) Adoption of artificial intelligence in breast imaging: evaluation, ethical constraints and limitations. Br J Cancer 125:15\u201322","journal-title":"Br J Cancer"},{"key":"2383_CR14","doi-asserted-by":"publisher","first-page":"6298","DOI":"10.1007\/s00330-024-10681-z","volume":"34","author":"MB Bergan","year":"2024","unstructured":"Bergan MB, Larsen M, Moshina N et al (2024) AI performance by mammographic density in a retrospective cohort study of 99,489 participants in BreastScreen Norway. Eur Radiol 34:6298\u20136308","journal-title":"Eur Radiol"},{"key":"2383_CR15","doi-asserted-by":"crossref","unstructured":"Marinovich ML, Wylie E, Lotter W et al (2023) Artificial intelligence (AI) for breast cancer screening: BreastScreen population-based cohort study of cancer detection. eBioMedicine 90:104498","DOI":"10.1016\/j.ebiom.2023.104498"},{"key":"2383_CR16","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.220146","volume":"5","author":"CF de Vries","year":"2023","unstructured":"de Vries CF, Colosimo SJ, Staff RT et al (2023) Impact of different mammography systems on artificial intelligence performance in breast cancer screening. Radiol Artif Intell 5:e220146","journal-title":"Radiol Artif Intell"},{"key":"2383_CR17","doi-asserted-by":"publisher","first-page":"e507","DOI":"10.1016\/S2589-7500(22)00070-X","volume":"4","author":"C Leibig","year":"2022","unstructured":"Leibig C, Brehmer M, Bunk S, Byng D, Pinker K, Umutlu L (2022) Combining the strengths of radiologists and AI for breast cancer screening: a retrospective analysis. Lancet Digit Health 4:e507\u2013e519","journal-title":"Lancet Digit Health"},{"key":"2383_CR18","doi-asserted-by":"publisher","DOI":"10.1186\/s12885-023-10890-7","volume":"23","author":"N Sharma","year":"2023","unstructured":"Sharma N, Ng AY, James JJ et al (2023) Multi-vendor evaluation of artificial intelligence as an independent reader for double reading in breast cancer screening on 275,900 mammograms. BMC Cancer 23:460","journal-title":"BMC Cancer"},{"key":"2383_CR19","doi-asserted-by":"publisher","first-page":"051807","DOI":"10.1117\/1.JMI.10.5.051807","volume":"10","author":"M Riveira-Martin","year":"2023","unstructured":"Riveira-Martin M, Rodriguez-Ruiz A, Marti R, Chevalier M (2023) Multi-vendor robustness analysis of a commercial artificial intelligence system for breast cancer detection. J Med Imaging 10:051807","journal-title":"J Med Imaging"},{"key":"2383_CR20","doi-asserted-by":"publisher","first-page":"826","DOI":"10.1148\/radiol.2015151516","volume":"277","author":"PM Bossuyt","year":"2015","unstructured":"Bossuyt PM, Reitsma JB, Bruns DE et al (2015) STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. Radiology 277:826\u2013832","journal-title":"Radiology"},{"key":"2383_CR21","doi-asserted-by":"publisher","first-page":"33","DOI":"10.1186\/s13244-025-01905-x","volume":"16","author":"E Kotter","year":"2025","unstructured":"Kotter E, D\u2019Antonoli TA, Cuocolo R et al (2025) Guiding AI in radiology: ESR\u2019s recommendations for effective implementation of the European AI Act. Insights Imaging 16:33","journal-title":"Insights Imaging"},{"key":"2383_CR22","doi-asserted-by":"publisher","first-page":"6145","DOI":"10.1007\/s00330-024-10661-3","volume":"34","author":"ME Seker","year":"2024","unstructured":"Seker ME, Koyluoglu YO, Ozaydin AN et al (2024) Diagnostic capabilities of artificial intelligence as an additional reader in a breast cancer screening program. Eur Radiol 34:6145\u20136157","journal-title":"Eur Radiol"},{"key":"2383_CR23","doi-asserted-by":"crossref","unstructured":"Lehnen T, Polenske D, Wichtmann BD, Lehnen NC (2026) AI software as a third reader in breast cancer screening\u2014a prospective diagnostic observational study. Eur Radiol 36:4478\u20134488","DOI":"10.1007\/s00330-026-12359-0"},{"key":"2383_CR24","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.230375","volume":"6","author":"M Larsen","year":"2024","unstructured":"Larsen M, Olstad CF, Lee CI et al (2024) Performance of an artificial intelligence system for breast cancer detection on screening mammograms from BreastScreen Norway. Radiol Artif Intell 6:e230375","journal-title":"Radiol Artif Intell"},{"key":"2383_CR25","doi-asserted-by":"publisher","first-page":"16","DOI":"10.1186\/s13244-023-01541-3","volume":"15","author":"AP Brady","year":"2024","unstructured":"Brady AP, Allen B, Chong J et al (2024) Developing, purchasing, implementing and monitoring AI tools in radiology: practical considerations. A multi-society statement from the ACR, CAR, ESR, RANZCR & RSNA. Insights Imaging 15:16","journal-title":"Insights Imaging"},{"key":"2383_CR26","doi-asserted-by":"crossref","unstructured":"Cai L, Golatta M, Sidey-Gibbons C, Barr RG, Pfob A (2025) The impact of updated imaging software on the performance of machine learning models for breast cancer diagnosis: a multi-center, retrospective study. Arch Gynecol Obstet 312:139\u2013147","DOI":"10.1007\/s00404-024-07901-8"},{"key":"2383_CR27","doi-asserted-by":"publisher","DOI":"10.7717\/peerj-cs.2476","volume":"11","author":"D Dave","year":"2025","unstructured":"Dave D, Akhunzada A, Ivkovic N et al (2025) Diagnostic test accuracy of AI-assisted mammography for breast imaging: a narrative review. PeerJ Comput Sci 11:e2476","journal-title":"PeerJ Comput Sci"},{"key":"2383_CR28","doi-asserted-by":"publisher","first-page":"e0000973","DOI":"10.1371\/journal.pdig.0000973","volume":"4","author":"AJ Hickman","year":"2025","unstructured":"Hickman AJ, Gomes S, Warren LM, Smith NAS, Shenton-Taylor C (2025) Assessing the generalisation of artificial intelligence across mammography manufacturers. PLoS Digit Health 4:e0000973","journal-title":"PLoS Digit Health"},{"key":"2383_CR29","first-page":"014001","volume":"11","author":"SD Verboom","year":"2024","unstructured":"Verboom SD, Caballo M, Peters J et al (2024) Deep learning-based breast region segmentation in raw and processed digital mammograms: generalization across views and vendors. J Med Imaging 11:014001","journal-title":"J Med Imaging"},{"key":"2383_CR30","doi-asserted-by":"publisher","first-page":"90","DOI":"10.1186\/s13244-023-01396-8","volume":"14","author":"P Hejduk","year":"2023","unstructured":"Hejduk P, Sexauer R, Ruppert C, Borkowski K, Unkelbach J, Schmidt N (2023) Automatic and standardized quality assurance of digital mammography and tomosynthesis with deep convolutional neural networks. Insights Imaging 14:90","journal-title":"Insights Imaging"},{"key":"2383_CR31","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-022-23990-4","volume":"12","author":"O Kilim","year":"2022","unstructured":"Kilim O, Olar A, Joo T, Palicz T, Pollner P, Csabai I (2022) Physical imaging parameter variation drives domain shift. Sci Rep 12:21302","journal-title":"Sci Rep"},{"key":"2383_CR32","first-page":"75","volume":"31","author":"B Kocak","year":"2025","unstructured":"Kocak B, Ponsiglione A, Stanzione A et al (2025) Bias in artificial intelligence for medical imaging: fundamentals, detection, avoidance, mitigation, challenges, ethics, and prospects. Diagn Interv Radiol 31:75\u201388","journal-title":"Diagn Interv Radiol"},{"key":"2383_CR33","doi-asserted-by":"publisher","first-page":"259","DOI":"10.1038\/s41746-025-01623-0","volume":"8","author":"H Schurz","year":"2025","unstructured":"Schurz H, Solander K, Astrom D et al (2025) Simulating mismatch between calibration and target population in AI for mammography the retrospective VAIB study. NPJ Digit Med 8:259","journal-title":"NPJ Digit Med"},{"key":"2383_CR34","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.222176","volume":"307","author":"T Dratsch","year":"2023","unstructured":"Dratsch T, Chen X, Rezazade Mehrizi M et al (2023) Automation bias in mammography: the impact of artificial intelligence BI-RADS suggestions on reader performance. Radiology 307:e222176","journal-title":"Radiology"},{"key":"2383_CR35","doi-asserted-by":"publisher","first-page":"471","DOI":"10.1007\/s10549-016-3882-0","volume":"158","author":"RJ Weber","year":"2016","unstructured":"Weber RJ, van Bommel RM, Louwman MW et al (2016) Characteristics and prognosis of interval cancers after biennial screen-film or full-field digital screening mammography. Breast Cancer Res Treat 158:471\u2013483","journal-title":"Breast Cancer Res Treat"}],"container-title":["Insights into Imaging"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s13244-026-02383-5.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s13244-026-02383-5","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s13244-026-02383-5.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T12:52:32Z","timestamp":1787748752000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1186\/s13244-026-02383-5"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,26]]},"references-count":35,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["2383"],"URL":"https:\/\/doi.org\/10.1186\/s13244-026-02383-5","relation":{},"ISSN":["1869-4101"],"issn-type":[{"value":"1869-4101","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,26]]},"assertion":[{"value":"8 December 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"3 July 2026","order":2,"name":"revised","label":"Revised","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"2 August 2026","order":3,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"26 August 2026","order":4,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"Institutional Review Board approval was obtained. The research study has been approved by the responsible regional ethics committee of Eastern Switzerland (EKOS 24\/132, ID 2024-01310). Participants in the screening program independently need to give their consent for using their anonymized data for statistical analyses. The study followed the STARD guideline for reporting diagnostic accuracy studies.","order":1,"name":"Ethics","label":"Ethics approval and consent to participate","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"Not applicable.","order":2,"name":"Ethics","label":"Consent for publication","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare that they have no competing interests.","order":3,"name":"Ethics","label":"Competing interests","group":{"name":"EthicsHeading","label":"Declarations"}}],"article-number":"209"}}