{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,13]],"date-time":"2026-10-13T06:05:52Z","timestamp":1791871552898,"version":"4.3.4"},"reference-count":28,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,6,15]],"date-time":"2026-06-15T00:00:00Z","timestamp":1781481600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,6,18]],"date-time":"2026-06-18T00:00:00Z","timestamp":1781740800000},"content-version":"vor","delay-in-days":3,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"the scientific Research Fund of Zhejiang Provincial Education Department","award":["Y202455079"],"award-info":[{"award-number":["Y202455079"]}]},{"DOI":"10.13039\/501100001809","name":"The National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["62576319"],"award-info":[{"award-number":["62576319"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["BMC Musculoskelet Disord"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>We developed a multi-modal Transformer framework integrating knee radiographs with clinical covariates to enable automated, objective, and generalizable ordinal Kellgren-Lawrence (KL) grading.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      A total of 2,703 anteroposterior knee radiographs were retrospectively collected from three independent medical centers (January 2018 - December 2024). Data from two centers (\n                      <jats:italic>n<\/jats:italic>\n                      \u2009=\u20091,953) were used for model development and internal five-fold stratified cross-validation, while the third center (\n                      <jats:italic>n<\/jats:italic>\n                      \u2009=\u2009750) served as an independent external test set. The proposed framework combines a Swin Transformer-Base image encoder with a clinical feature Transformer through a novel Robust Cross-Modal Gated Fusion (RCGF) module employing bidirectional cross-attention and uncertainty-aware dynamic gating via Monte-Carlo dropout. Ordinal prediction was performed using Consistent Rank Logits (CORAL). Eight classifier architectures were systematically compared, encompassing multi-modal models, unimodal image-only baselines, and a clinical-only model.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      The proposed RCGF framework achieved a Quadratic Weighted Kappa (QWK) of 0.900 (95% CI: 0.877\u20130.921), macro-averaged AUC of 0.930 (95% CI: 0.910\u20130.950), and balanced accuracy of 87.6% on the independent external test set, significantly outperforming all baseline models including BioViL-T (QWK\u2009=\u20090.850) and MedViT (QWK\u2009=\u20090.830; all FDR-corrected\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001). Sensitivity for severe Osteoarthritis (OA) (Grade 4) reached 83.5% (95% CI: 79.1\u201387.4%), with specificity 95.3%. The clinical nomogram demonstrated excellent calibration (calibration slope\u2009=\u20090.98, Brier score\u2009=\u20090.072, C-statistic\u2009=\u20090.940) and superior net benefit over treat-all and treat-none strategies across all clinically relevant decision thresholds.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>This multi-modal Transformer framework with uncertainty-aware gated fusion provides robust external generalizability for ordinal knee OA severity grading and delivers a clinically actionable nomogram. The approach has strong potential to reduce radiologist workload and facilitate objective assessment on routine clinical radiographs, particularly in resource-constrained settings.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1186\/s12891-026-10043-4","type":"journal-article","created":{"date-parts":[[2026,6,15]],"date-time":"2026-06-15T07:16:14Z","timestamp":1781507774000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Clinically integrated multi-modal transformer framework with cross-modal gated fusion and clinical nomogram for automated Kellgren-Lawrence grading of knee osteoarthritis on x-ray images"],"prefix":"10.1186","volume":"27","author":[{"given":"Yingying","family":"Huang","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zihan","family":"Shao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Renfang","family":"Wang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hong","family":"Qiu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,6,15]]},"reference":[{"issue":"11","key":"10043_CR1","doi-asserted-by":"publisher","first-page":"1300","DOI":"10.1016\/j.joca.2025.08.015","volume":"33","author":"A Dell\u2019Isola","year":"2025","unstructured":"Dell\u2019Isola A, Recenti F, Giardulli B, Lawford BJ, Kiadaliri A. Osteoarthritis year in review 2025: Epidemiology and therapy. Osteoarthritis Cartilage. 2025;33(11):1300\u20136.","journal-title":"Osteoarthritis Cartilage"},{"issue":"11","key":"10043_CR2","doi-asserted-by":"publisher","first-page":"1397","DOI":"10.1016\/j.joca.2024.07.014","volume":"32","author":"A Courties","year":"2024","unstructured":"Courties A, Kouki I, Soliman N, Mathieu S, Sellam J. Osteoarthritis year in review 2024: Epidemiology and therapy. Osteoarthritis Cartilage. 2024;32(11):1397\u2013404.","journal-title":"Osteoarthritis Cartilage"},{"issue":"8","key":"10043_CR3","doi-asserted-by":"publisher","first-page":"1886","DOI":"10.1007\/s11999-016-4732-4","volume":"474","author":"MD Kohn","year":"2016","unstructured":"Kohn MD, Sassoon AA, Fernando ND. Classifications in Brief: Kellgren-Lawrence Classification of Osteoarthritis. Clin Orthop Relat Res. 2016;474(8):1886\u201393.","journal-title":"Clin Orthop Relat Res"},{"issue":"2","key":"10043_CR4","doi-asserted-by":"publisher","first-page":"100580","DOI":"10.1016\/j.ocarto.2025.100580","volume":"7","author":"E Vaattovaara","year":"2025","unstructured":"Vaattovaara E, Panfilov E, Tiulpin A, Niinim\u00e4ki T, Niinim\u00e4ki J, Saarakkala S, Nevalainen MT. Kellgren-Lawrence grading of knee osteoarthritis using deep learning: Diagnostic performance with external dataset and comparison with four readers. Osteoarthr Cartil Open. 2025;7(2):100580.","journal-title":"Osteoarthr Cartil Open"},{"issue":"1","key":"10043_CR5","doi-asserted-by":"publisher","first-page":"3","DOI":"10.3390\/surgeries6010003","volume":"6","author":"D Nasef","year":"2025","unstructured":"Nasef D, Nasef D, Sawiris V, Girgis P, Toma M. Deep Learning for Automated Kellgren\u2013Lawrence Grading in Knee Osteoarthritis Severity Assessment. Surgeries. 2025;6(1):3.","journal-title":"Surgeries"},{"key":"10043_CR6","doi-asserted-by":"crossref","unstructured":"Lavanya RSS. Deep learning in knee osteoarthritis: A task-based systematic review of recent advances. Biomed Signal Process Control. 2026;119:109771.","DOI":"10.1016\/j.bspc.2026.109771"},{"issue":"4","key":"10043_CR7","doi-asserted-by":"publisher","first-page":"e70151","DOI":"10.1002\/ima.70151","volume":"35","author":"S Jafarpoor Nesheli","year":"2025","unstructured":"Jafarpoor Nesheli S, Sabet M, Firoozi V, Heydarheydari S, Rezaeijo SM. Enhanced Interpretability in Breast Cancer Detection: Combining Grad-CAM With Selective Layer Freezing in Deep Learning. Int J Imaging Syst Technol. 2025;35(4):e70151.","journal-title":"Int J Imaging Syst Technol"},{"issue":"6","key":"10043_CR8","doi-asserted-by":"publisher","first-page":"963","DOI":"10.3390\/math14060963","volume":"14","author":"Z Ullah","year":"2026","unstructured":"Ullah Z, Kim J. VL-OrdinalFormer: Vision\u2013Language-Guided Ordinal Transformers for Interpretable Knee Osteoarthritis Grading. Mathematics. 2026;14(6):963.","journal-title":"Mathematics"},{"key":"10043_CR9","doi-asserted-by":"crossref","unstructured":"Ren X, Hou L, Liu S, Wu P, Liang S, Fu H et al. OA-MEN: a fusion deep learning approach for enhanced accuracy in knee osteoarthritis detection and classification using X-Ray imaging. Front Bioeng Biotechnol. 2025;12:1437188.","DOI":"10.3389\/fbioe.2024.1437188"},{"issue":"1","key":"10043_CR10","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s13534-024-00437-5","volume":"15","author":"T Tariq","year":"2025","unstructured":"Tariq T, Suhail Z, Nawaz Z. A Review for automated classification of knee osteoarthritis using KL grading scheme for X-rays. Biomed Eng Lett. 2025;15(1):1\u201335.","journal-title":"Biomed Eng Lett"},{"issue":"1","key":"10043_CR11","doi-asserted-by":"publisher","first-page":"194","DOI":"10.1007\/s12672-025-04359-1","volume":"17","author":"S Salehzehi","year":"2025","unstructured":"Salehzehi S, Jafarpoor Nesheli S, Eftekhar A, Shojaee Rocky M, Vejdani Noghreiyan V, Bazazpour A, et al. Graph neural network-based multi-label classification of breast cancer using multiparametric MRI and clinical features. Discover Oncol. 2025;17(1):194.","journal-title":"Discover Oncol"},{"key":"10043_CR12","doi-asserted-by":"publisher","first-page":"109056","DOI":"10.1016\/j.cmpb.2025.109056","volume":"272","author":"SM Rezaeijo","year":"2025","unstructured":"Rezaeijo SM, Eftekhar A, Rouhi S, Keshavarzi B, Mohammadi Z, Firouzabad LA, et al. Neighboring tissues as diagnostic windows: Neighborhood effects in radiomic detection of pancreatic ductal adenocarcinoma. Comput Methods Programs Biomed. 2025;272:109056.","journal-title":"Comput Methods Programs Biomed"},{"key":"10043_CR13","doi-asserted-by":"publisher","first-page":"106791","DOI":"10.1016\/j.compbiomed.2023.106791","volume":"157","author":"ON Manzari","year":"2023","unstructured":"Manzari ON, Ahmadabadi H, Kashiani H, Shokouhi SB, Ayatollahi A. MedViT A Robust Vision Transformer for Generalized Medical Image Classification. Comput Biol Med. 2023;157:106791.","journal-title":"Comput Biol Med"},{"key":"10043_CR14","doi-asserted-by":"crossref","unstructured":"Xu J, Lei M, Xu D. Global, regional, and national burden of knee osteoarthritis attributable to high BMI: a systematic analysis from 1990 to 2021 and projections to 2050. Front Public Health. 2025;13:1668257.","DOI":"10.3389\/fpubh.2025.1668257"},{"key":"10043_CR15","first-page":"18932","volume":"37","author":"Y Gorishniy","year":"2024","unstructured":"Gorishniy Y, Rubachev I, Khrulkov V, Babenko A. Revisiting Deep Learning Models for Tabular Data. Adv Neural Inf Process Syst. 2024;37:18932\u201343.","journal-title":"Adv Neural Inf Process Syst"},{"key":"10043_CR16","doi-asserted-by":"crossref","unstructured":"Du G, Zeng X, Wu R, Xu J. Uncertainty-based data augmentation and pseudo-label refinement for semi-supervised medical image segmentation. Displays. 2026;94:103486.","DOI":"10.1016\/j.displa.2026.103486"},{"key":"10043_CR17","doi-asserted-by":"crossref","unstructured":"Zeng X, Xiong S, Xu J, Du G, Rong Y. Uncertainty co-estimator for improving semi-supervised medical image segmentation. IEEE Trans Med Imaging. 2025;44(9):3870\u201381.","DOI":"10.1109\/TMI.2025.3570310"},{"key":"10043_CR18","unstructured":"Barekatain L, Glocker B. Evaluating the explainability of vision transformers in medical imaging. [Preprint] arXiv:2510.12021 [cs.CV]. 2025. https:\/\/arxiv.org\/abs\/2510.12021."},{"key":"10043_CR19","unstructured":"Fang C, Sandino C, Mahasseni B, Minxha J, Pouransari H, Azemi E, et al. Promoting cross-modal representations to improve multimodal foundation models for physiological signals. [Preprint] arXiv:2410.16424 [cs.LG]. 2024. https:\/\/arxiv.org\/abs\/2410.16424."},{"key":"10043_CR20","doi-asserted-by":"crossref","unstructured":"Bannur S, Hyland S, Liu Q, Perez-Garcia F, Ilse M, Castro DC, Boecking B, editors. Learning to Exploit Temporal Structure for Biomedical Vision-Language Processing. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 2023.","DOI":"10.1109\/CVPR52729.2023.01442"},{"key":"10043_CR21","doi-asserted-by":"crossref","unstructured":"Chowdary GJ, Yin Z, editors. Med-former: A transformer based architecture for medical image classification. International conference on medical image computing and computer-assisted intervention; 2024: Springer.","DOI":"10.1007\/978-3-031-72120-5_42"},{"key":"10043_CR22","unstructured":"Kim J, Lee S, Park H. CBAM-EfficientNetV2 for Histopathology Image Classification. IEEE Access. 2025."},{"key":"10043_CR23","doi-asserted-by":"crossref","unstructured":"P\u0142otka S, Chrabaszcz M, Biecek P, editors. Swin smt: Global sequential modeling for enhancing 3d medical image segmentation. International Conference on Medical Image Computing and Computer-Assisted Intervention; 2024: Springer.","DOI":"10.1007\/978-3-031-72111-3_65"},{"issue":"1","key":"10043_CR24","doi-asserted-by":"publisher","first-page":"22887","DOI":"10.1038\/s41598-023-50210-4","volume":"13","author":"S-W Pi","year":"2023","unstructured":"Pi S-W, Lee B-D, Lee MS, Lee HJ. Ensemble deep-learning networks for automated osteoarthritis grading in knee X-ray images. Sci Rep. 2023;13(1):22887.","journal-title":"Sci Rep"},{"issue":"1","key":"10043_CR25","doi-asserted-by":"publisher","first-page":"e70254","DOI":"10.1111\/nyas.70254","volume":"1557","author":"T Alkhatatbeh","year":"2026","unstructured":"Alkhatatbeh T, Alkhatatbeh A, Liao Y, Zhang Z, Fang H, Chen W, Zhang R. Cross-Institutional Five-Class Kellgren-Lawrence Grading of Knee Osteoarthritis via Multitask Deep Learning. Ann N Y Acad Sci. 2026;1557(1):e70254.","journal-title":"Ann N Y Acad Sci"},{"issue":"3","key":"10043_CR26","first-page":"608","volume":"33","author":"MKS Ma\u2019aitah","year":"2025","unstructured":"Ma\u2019aitah MKS, Helwan A, Radwan A, Manasreh AMS, Alshareef EA. Multimodal model for knee osteoarthritis KL grading from plain radiograph. J X-Ray Sci Technol. 2025;33(3):608\u201320.","journal-title":"J X-Ray Sci Technol"},{"key":"10043_CR27","doi-asserted-by":"publisher","first-page":"196654","DOI":"10.1109\/ACCESS.2025.3634156","volume":"13","author":"M Hayat","year":"2025","unstructured":"Hayat M, Aramvith S. Superpixel-Guided Graph-Attention Boundary GAN for Adaptive Feature Refinement in Scribble-Supervised Medical Image Segmentation. IEEE Access. 2025;13:196654\u201368.","journal-title":"IEEE Access"},{"key":"10043_CR28","doi-asserted-by":"crossref","unstructured":"Hayat M, Dhaliwal A, Din MMU, Izhar R, Nadeem M, Ahmad N, editors. Cross-Attention Patch Fusion for Few-Shot Colorectal Tissue Generation. 2025 5th International Conference on Digital Futures and Transformative Technologies (ICoDT2); 2025: IEEE.","DOI":"10.1109\/ICoDT269104.2025.11360745"}],"container-title":["BMC Musculoskeletal Disorders"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s12891-026-10043-4","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s12891-026-10043-4.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s12891-026-10043-4.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,18]],"date-time":"2026-06-18T15:18:07Z","timestamp":1781795887000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1186\/s12891-026-10043-4"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,15]]},"references-count":28,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["10043"],"URL":"https:\/\/doi.org\/10.1186\/s12891-026-10043-4","relation":{},"ISSN":["1471-2474"],"issn-type":[{"value":"1471-2474","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,15]]},"assertion":[{"value":"1 April 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"1 June 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"15 June 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"This study was conducted as a retrospective analysis using de-identified clinical and imaging data. The study protocol was reviewed and approved by the Institutional Review Board of Zhejiang Wanli University, China (approval no. 62576319). The requirement for informed consent was waived by the Institutional Review Board due to the retrospective design of the study and the use of anonymized data. All procedures involving human participants were conducted in accordance with the ethical standards of the institutional research committee and with the 1964 Declaration of Helsinki and its later amendments.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval and consent to participate"}},{"value":"Not applicable.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent for publication"}},{"value":"The authors declare no competing interests.","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"528"}}