{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T00:04:42Z","timestamp":1789430682108,"version":"build-2803163510"},"reference-count":32,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,6,27]],"date-time":"2026-06-27T00:00:00Z","timestamp":1782518400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T00:00:00Z","timestamp":1789430400000},"content-version":"vor","delay-in-days":80,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["BMC Med Imaging"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>A Transformer-based multimodal deep learning model was developed to enhance ultrasound imaging diagnosis of PTC and benign nodules.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      This study included 491 thyroid nodule patients from two centers, with 406 from the Suzhou Ninth People\u2019s Hospital divided into training (\n                      <jats:italic>n<\/jats:italic>\n                      \u2009=\u2009284) and validation (\n                      <jats:italic>n<\/jats:italic>\n                      \u2009=\u2009122) sets, and 85 from Jiangsu Shengze Hospital as an external test set. The dataset comprised 232 benign nodules and 259 papillary thyroid carcinoma cases. A comparison of 34 deep learning architectures and four traditional models was conducted, leading to the proposal of an Efficientnetv2scbamTrans fusion model. Five ablation experiments assessed module contributions, while SHAP and Grad-CAM were used for interpretability, with energy concentration, peak center distance, and spatial IoU as performance indicators.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      DenseNet201 excelled on the internal validation set but showed overfitting with an AUC of 0.702 on the external test set. The unimodal imaging baseline had an AUC of 0.782, which improved to 0.811 with clinical features, reducing overfitting risk. Directly fusing radiomics features yielded no improvement, maintaining an AUC of 0.751. The proposed Efficientnetv2scbamTrans model reached an AUC of 0.985 on the internal validation set. More importantly, on the independent external test set (Jiangsu Shengze Hospital), the model achieved an AUC of 0.986 (95% CI: 0.967\u20131.000), with 96.4% sensitivity and 96.7% specificity. On this external test set, our proposed model significantly outperformed the traditional DLR model by an AUC margin of 0.199 (\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001). SHAP analysis indicated clinical features altered radiomics decision weights, and Grad-CAM showed an increase in spatial IoU from 0.097 to 0.349, enhancing visual localization.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>The study developed a multimodal deep learning model using thyroid ultrasound, improving diagnostics with interaction and attention mechanisms via CBAM and Transformer.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1186\/s12880-026-02530-w","type":"journal-article","created":{"date-parts":[[2026,6,29]],"date-time":"2026-06-29T05:58:13Z","timestamp":1782712693000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Application of transformer attention mechanism-based multimodal deep learning model in the diagnosis of papillary thyroid carcinoma"],"prefix":"10.1186","volume":"26","author":[{"given":"Min","family":"Xu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xufeng","family":"Zhu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhaoyang","family":"Lu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Weihua","family":"Xu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chunming","family":"Shen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7763-6571","authenticated-orcid":false,"given":"Jing","family":"Yang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,6,27]]},"reference":[{"key":"2530_CR1","doi-asserted-by":"publisher","unstructured":"Lebbink CA, Links TP, Czarniecka A, Dias RP, Elisei R, Izatt L, et al. 2022 European Thyroid Association guidelines for the management of pediatric thyroid nodules and differentiated thyroid carcinoma. Eur Thyroid J. 2022;11. https:\/\/doi.org\/10.1530\/ETJ-22-0146.","DOI":"10.1530\/ETJ-22-0146"},{"key":"2530_CR2","doi-asserted-by":"publisher","first-page":"92","DOI":"10.5694\/mja17.01204","volume":"209","author":"R Wong","year":"2018","unstructured":"Wong R, Farrell SG, Grossmann M. Thyroid nodules: diagnosis and management. Med J Australia. 2018;209:92\u20138. https:\/\/doi.org\/10.5694\/mja17.01204.","journal-title":"Med J Australia"},{"key":"2530_CR3","doi-asserted-by":"publisher","first-page":"914","DOI":"10.1001\/jama.2018.0898","volume":"319","author":"C Durante","year":"2018","unstructured":"Durante C, Grani G, Lamartina L, Filetti S, Mandel SJ, Cooper DS. The Diagnosis and Management of Thyroid Nodules: A Review. Jama-J Am Med Assoc. 2018;319:914\u201324. https:\/\/doi.org\/10.1001\/jama.2018.0898.","journal-title":"Jama-J Am Med Assoc"},{"key":"2530_CR4","doi-asserted-by":"publisher","first-page":"1253","DOI":"10.1210\/jc.2013-2928","volume":"99","author":"JP Brito","year":"2014","unstructured":"Brito JP, Gionfriddo MR, Al Nofal A, Boehmer KR, Leppin AL, Reading C, et al. The Accuracy of Thyroid Nodule Ultrasound to Predict Thyroid Cancer: Systematic Review and Meta-Analysis. J Clin Endocrinol Metabolism. 2014;99:1253\u201363. https:\/\/doi.org\/10.1210\/jc.2013-2928.","journal-title":"J Clin Endocrinol Metabolism"},{"key":"2530_CR5","doi-asserted-by":"publisher","first-page":"162","DOI":"10.2214\/AJR.17.19192","volume":"211","author":"JK Hoang","year":"2018","unstructured":"Hoang JK, Middleton WD, Farjat AE, Teefey SA, Abinanti N, Boschini FJ, et al. Interobserver Variability of Sonographic Features Used in the American College of Radiology Thyroid Imaging Reporting and Data System. Am J Roentgenol. 2018;211:162\u20137. https:\/\/doi.org\/10.2214\/AJR.17.19192.","journal-title":"Am J Roentgenol"},{"key":"2530_CR6","doi-asserted-by":"publisher","first-page":"173","DOI":"10.1007\/s13244-015-0398-9","volume":"6","author":"AS McQueen","year":"2015","unstructured":"McQueen AS, Bhatia KS. Thyroid nodule ultrasound: technical advances and future horizons. Insights Imaging. 2015;6:173\u201388. https:\/\/doi.org\/10.1007\/s13244-015-0398-9.","journal-title":"Insights Imaging"},{"key":"2530_CR7","doi-asserted-by":"publisher","first-page":"744","DOI":"10.1007\/s12020-024-04053-2","volume":"87","author":"Z Chen","year":"2025","unstructured":"Chen Z, Chambara N, Lo X, Liu S, Gunda ST, Han X, et al. Improving the diagnostic strategy for thyroid nodules: a combination of artificial intelligence-based computer-aided diagnosis system and shear wave elastography. Endocrine. 2025;87:744\u201357. https:\/\/doi.org\/10.1007\/s12020-024-04053-2.","journal-title":"Endocrine"},{"key":"2530_CR8","doi-asserted-by":"publisher","first-page":"209","DOI":"10.1102\/1470-7330.2011.0030","volume":"11","author":"G Anil","year":"2011","unstructured":"Anil G, Hegde A, Chong FH. Thyroid nodules: risk stratification for malignancy with ultrasound and guided biopsy. Cancer Imaging. 2011;11:209\u201323. https:\/\/doi.org\/10.1102\/1470-7330.2011.0030.","journal-title":"Cancer Imaging"},{"key":"2530_CR9","doi-asserted-by":"publisher","DOI":"10.7759\/cureus.7239","author":"R Mistry","year":"2020","unstructured":"Mistry R, Hillyar C, Nibber A, Sooriyamoorthy T, Kumar N. Ultrasound Classification of Thyroid Nodules: A Systematic Review. Cureus J Med Sci. 2020. https:\/\/doi.org\/10.7759\/cureus.7239.","journal-title":"Cureus J Med Sci"},{"key":"2530_CR10","doi-asserted-by":"publisher","first-page":"1321118","DOI":"10.3389\/fsurg.2023.1321118","volume":"10","author":"L Zhang","year":"2023","unstructured":"Zhang L, Wan R, Chen J, Xin F, Han H. Analysis of the correlation between clinical and imaging features of malignant lung nodules and pathological types. Front Surg. 2023;10:1321118. https:\/\/doi.org\/10.3389\/fsurg.2023.1321118.","journal-title":"Front Surg"},{"key":"2530_CR11","doi-asserted-by":"publisher","first-page":"1","DOI":"10.5604\/01.3001.0014.3433","volume":"75","author":"P Singh","year":"2020","unstructured":"Singh P, Gupta N, Dass A, Handa U, Singhal S. Correlation of fine needle aspiration cytology with histopathology in patients undergoing thyroid surgery. Pol J Otolaryngol. 2020;75:1\u20135. https:\/\/doi.org\/10.5604\/01.3001.0014.3433.","journal-title":"Pol J Otolaryngol"},{"key":"2530_CR12","doi-asserted-by":"publisher","first-page":"954","DOI":"10.1038\/s41591-019-0447-x","volume":"25","author":"D Ardila","year":"2019","unstructured":"Ardila D, Kiraly AP, Bharadwaj S, Choi B, Reicher JJ, Peng L, et al. End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nat Med. 2019;25:954\u201361. https:\/\/doi.org\/10.1038\/s41591-019-0447-x.","journal-title":"Nat Med"},{"key":"2530_CR13","doi-asserted-by":"publisher","first-page":"193","DOI":"10.1016\/S1470-2045(18)30762-9","volume":"20","author":"X Li","year":"2019","unstructured":"Li X, Zhang S, Zhang Q, Wei X, Pan Y, Zhao J, et al. Diagnosis of thyroid cancer using deep convolutional neural network models applied to sonographic images: a retrospective, multicohort, diagnostic study. Lancet Oncol. 2019;20:193\u2013201. https:\/\/doi.org\/10.1016\/S1470-2045(18)30762-9.","journal-title":"Lancet Oncol"},{"key":"2530_CR14","doi-asserted-by":"publisher","first-page":"695","DOI":"10.1148\/radiol.2019181343","volume":"292","author":"M Buda","year":"2019","unstructured":"Buda M, Wildman-Tobriner B, Hoang JK, Thayer D, Tessler FN, Middleton WD, et al. Management of Thyroid Nodules Seen on US Images: Deep Learning May Match Performance of Radiologists. Radiology. 2019;292:695\u2013701. https:\/\/doi.org\/10.1148\/radiol.2019181343.","journal-title":"Radiology"},{"key":"2530_CR15","doi-asserted-by":"publisher","first-page":"221","DOI":"10.1016\/j.ultras.2016.09.011","volume":"73","author":"J Ma","year":"2017","unstructured":"Ma J, Wu F, Zhu J, Xu D, Kong D. A pre-trained convolutional neural network based method for thyroid nodule diagnosis. Ultrasonics. 2017;73:221\u201330. https:\/\/doi.org\/10.1016\/j.ultras.2016.09.011.","journal-title":"Ultrasonics"},{"key":"2530_CR16","doi-asserted-by":"publisher","first-page":"1303","DOI":"10.7150\/thno.30309","volume":"9","author":"Z Liu","year":"2019","unstructured":"Liu Z, Wang S, Dong D, Wei J, Fang C, Zhou X, et al. The Applications of Radiomics in Precision Diagnosis and Treatment of Oncology: Opportunities and Challenges. Theranostics. 2019;9:1303\u201322. https:\/\/doi.org\/10.7150\/thno.30309.","journal-title":"Theranostics"},{"key":"2530_CR17","doi-asserted-by":"publisher","first-page":"749","DOI":"10.1038\/nrclinonc.2017.141","volume":"14","author":"P Lambin","year":"2017","unstructured":"Lambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, et al. Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol. 2017;14:749\u201362. https:\/\/doi.org\/10.1038\/nrclinonc.2017.141.","journal-title":"Nat Rev Clin Oncol"},{"key":"2530_CR18","doi-asserted-by":"publisher","first-page":"26007","DOI":"10.1088\/1741-2552\/aaf3f6","volume":"16","author":"F Fahimi","year":"2019","unstructured":"Fahimi F, Zhang Z, Goh WB, Lee TS, Ang KK, Guan C. Inter-subject transfer learning with an end-to-end deep convolutional neural network for EEG-based BCI. J Neural Eng. 2019;16:26007. https:\/\/doi.org\/10.1088\/1741-2552\/aaf3f6.","journal-title":"J Neural Eng"},{"key":"2530_CR19","doi-asserted-by":"crossref","unstructured":"Woo S, Park J, Lee J, Kweon IS. CBAM: Convolutional block attention module. In: Ferrari V, Hebert M, Sminchisescu C, Weiss Y, editors. Computer Vision \u2013 ECCV 2018 Workshops. Cham: Springer International Publishing; 2018. p. 3\u201319.","DOI":"10.1007\/978-3-030-01234-2_1"},{"key":"2530_CR20","doi-asserted-by":"publisher","first-page":"285","DOI":"10.1016\/j.nic.2021.04.001","volume":"31","author":"HS Malhi","year":"2021","unstructured":"Malhi HS, Grant EG. Ultrasound of Thyroid Nodules and the Thyroid Imaging Reporting and Data System. Neuroimag Clin N Am. 2021;31:285\u2013300. https:\/\/doi.org\/10.1016\/j.nic.2021.04.001.","journal-title":"Neuroimag Clin N Am"},{"key":"2530_CR21","doi-asserted-by":"publisher","first-page":"1109319","DOI":"10.3389\/fonc.2023.1109319","volume":"13","author":"X Gao","year":"2023","unstructured":"Gao X, Ran X, Ding W. The progress of radiomics in thyroid nodules. Front Oncol. 2023;13:1109319. https:\/\/doi.org\/10.3389\/fonc.2023.1109319.","journal-title":"Front Oncol"},{"key":"2530_CR22","doi-asserted-by":"publisher","first-page":"5932","DOI":"10.21037\/qims-23-1597","volume":"14","author":"H Du","year":"2024","unstructured":"Du H, Chen F, Li H, Wang K, Zhang J, Meng J, et al. Deep-learning radiomics based on ultrasound can objectively evaluate thyroid nodules and assist in improving the diagnostic level of ultrasound physicians. Quant Imag Med Surg. 2024;14:5932\u201345. https:\/\/doi.org\/10.21037\/qims-23-1597.","journal-title":"Quant Imag Med Surg"},{"key":"2530_CR23","doi-asserted-by":"publisher","first-page":"648","DOI":"10.3390\/bioengineering11070648","volume":"11","author":"S Vahdati","year":"2024","unstructured":"Vahdati S, Khosravi B, Robinson KA, Rouzrokh P, Moassefi M, Akkus Z, et al. A Multi-View Deep Learning Model for Thyroid Nodules Detection and Characterization in Ultrasound Imaging. Bioengineering. 2024;11:648. https:\/\/doi.org\/10.3390\/bioengineering11070648.","journal-title":"Bioengineering"},{"key":"2530_CR24","doi-asserted-by":"publisher","first-page":"544","DOI":"10.1186\/s12885-025-13917-3","volume":"25","author":"Y Xu","year":"2025","unstructured":"Xu Y, Xu M, Geng Z, Liu J, Meng B. Thyroid nodule classification in ultrasound imaging using deep transfer learning. BMC Cancer. 2025;25:544. https:\/\/doi.org\/10.1186\/s12885-025-13917-3.","journal-title":"BMC Cancer"},{"key":"2530_CR25","doi-asserted-by":"publisher","first-page":"1771","DOI":"10.2217\/fon.10.127","volume":"6","author":"R Rahbari","year":"2010","unstructured":"Rahbari R, Zhang L, Kebebew E. Thyroid Cancer Gender Disparity. Future Oncol. 2010;6:1771\u20139. https:\/\/doi.org\/10.2217\/fon.10.127.","journal-title":"Future Oncol"},{"key":"2530_CR26","doi-asserted-by":"publisher","first-page":"5352","DOI":"10.3390\/jcm14155352","volume":"14","author":"N Angelopoulos","year":"2025","unstructured":"Angelopoulos N, Androulakis I, Askitis DP, Valvis N, Paparodis RD, Petkova V, et al. Integrating Clinical Parameters into Thyroid Nodule Malignancy Risk: A Retrospective Evaluation Based on ACR TI-RADS. J Clin Med. 2025;14:5352. https:\/\/doi.org\/10.3390\/jcm14155352.","journal-title":"J Clin Med"},{"key":"2530_CR27","doi-asserted-by":"publisher","first-page":"2436","DOI":"10.3390\/cancers13102436","volume":"13","author":"Y Cao","year":"2021","unstructured":"Cao Y, Zhong X, Diao W, Mu J, Cheng Y, Jia Z. Radiomics in Differentiated Thyroid Cancer and Nodules: Explorations, Application, and Limitations. Cancers. 2021;13:2436. https:\/\/doi.org\/10.3390\/cancers13102436.","journal-title":"Cancers"},{"key":"2530_CR28","unstructured":"Tan M, Le QV. EfficientNetV2: smaller models and faster training. 2021."},{"key":"2530_CR29","unstructured":"Zhang C, Kleinberg J, Raghu M, Bengio S. Transfusion: understanding transfer learning for medical imaging. Adv Neural Inf Process Syst. 2019."},{"key":"2530_CR30","doi-asserted-by":"publisher","first-page":"563","DOI":"10.1148\/radiol.2015151169","volume":"278","author":"RJ Gillies","year":"2016","unstructured":"Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278:563\u201377. https:\/\/doi.org\/10.1148\/radiol.2015151169.","journal-title":"Radiology"},{"key":"2530_CR31","doi-asserted-by":"crossref","unstructured":"Zwanenburg A, Leger S, Agolli L, Pilz K, Troost EGC, Richter C, et al. Assessing robustness of radiomic features by image perturbation. Sci Rep. 2019;9.","DOI":"10.1038\/s41598-018-36938-4"},{"key":"2530_CR32","doi-asserted-by":"publisher","first-page":"f4706","DOI":"10.1136\/bmj.f4706","volume":"347","author":"JP Brito","year":"2013","unstructured":"Brito JP, Morris JC, Montori VM. Thyroid cancer: zealous imaging has increased detection and treatment of low risk tumours. Bmj-Brit Med J. 2013;347:f4706. https:\/\/doi.org\/10.1136\/bmj.f4706.","journal-title":"Bmj-Brit Med J"}],"container-title":["BMC Medical Imaging"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s12880-026-02530-w","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s12880-026-02530-w.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s12880-026-02530-w.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T23:55:50Z","timestamp":1789430150000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1186\/s12880-026-02530-w"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,27]]},"references-count":32,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["2530"],"URL":"https:\/\/doi.org\/10.1186\/s12880-026-02530-w","relation":{},"ISSN":["1471-2342"],"issn-type":[{"value":"1471-2342","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,27]]},"assertion":[{"value":"31 January 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"23 June 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"27 June 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"This retrospective observational study used data from previously anonymized medical records and posed no additional risks to patients. Therefore, the Ethics Committee of Suzhou Ninth People\u2019s Hospital and Jiangsu Shengze Hospital approved a waiver for informed consent after they evaluated the study (Approval No. KYLW2025-083-01 and 2024-030-01). All data were analyzed in an anonymized form, adhering to the principles of the Declaration of Helsinki.","order":1,"name":"Ethics","label":"Ethical approval","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 no competing interests.","order":3,"name":"Ethics","label":"Competing interests","group":{"name":"EthicsHeading","label":"Declarations"}}],"article-number":"450"}}