{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,3]],"date-time":"2026-11-03T06:29:49Z","timestamp":1793687389377,"version":"4.5.1"},"reference-count":35,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T00:00:00Z","timestamp":1784160000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T00:00:00Z","timestamp":1784160000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Misr University for Science & Technology"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Sci Rep"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Accurate differentiation between benign and malignant thyroid nodules on ultrasound remains clinically important, yet interpretation is operator-dependent and subject to inter-observer variability. Recent deep learning studies report strong performance for thyroid ultrasound classification, but many prior approaches remain centred on image-level prediction, with limited emphasis on patient-level baselines, calibration-aware evaluation, and mask-related shortcut analysis. To address these gaps, we present a patient-level thyroid ultrasound malignancy prediction framework centred on a Region-Aware Context-Aware Fusion (RCAF) model evaluated under a strict leakage-free protocol. RCAF combines lesion-focused and context-preserving frame representations through a dual-branch design with gated fusion, followed by attention-based multiple instance learning (AttnMIL) for patient-level aggregation. The framework incorporates development-only probability calibration and threshold selection before single-shot evaluation on an untouched independent test cohort. Experiments on the public ThyroidXL benchmark show that RCAF outperforms stronger fair patient-level baselines, including image-only, transformer-based, and lesion-only comparators. Calibration analysis improves probability reliability, threshold analysis demonstrates stable behaviour under clinically relevant operating conditions, and shortcut sensitivity experiments show that naive mask concatenation produces shortcut-prone gains, whereas RCAF degrades by only 0.001 ROC-AUC under within-patient mask permutation, supporting principled region-aware reasoning. For cross-domain assessment, RCAF was evaluated on TN5000, a Chinese thyroid ultrasound dataset acquired under different imaging conditions; following domain adaptation and 8-view test-time augmentation, the model achieved AUC\u00a0=\u00a00.914 [0.879\u20130.948] on a class-balanced validation subset. Overall, RCAF constitutes a strong patient-level thyroid ultrasound classification framework, with encouraging cross-domain adaptability. Broader prospective multi-centre validation remains necessary before clinical deployment.<\/jats:p>","DOI":"10.1038\/s41598-026-61342-8","type":"journal-article","created":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T14:04:50Z","timestamp":1784210690000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["RCAF for patient-level thyroid ultrasound malignancy prediction under leakage-free evaluation and calibration"],"prefix":"10.1038","volume":"16","author":[{"given":"Mennatallah","family":"Sherif","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Eman K.","family":"Elsayed","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mohanad A.","family":"Deif","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,7,16]]},"reference":[{"issue":"4","key":"61342_CR1","doi-asserted-by":"publisher","first-page":"250","DOI":"10.1016\/S2589-7500(21)00041-8","volume":"3","author":"S Peng","year":"2021","unstructured":"Peng, S. et al. Deep learning-based artificial intelligence model to assist thyroid nodule diagnosis and management: A multicentre diagnostic study. Lancet Digit. Health 3(4), 250\u2013259. https:\/\/doi.org\/10.1016\/S2589-7500(21)00041-8 (2021).","journal-title":"Lancet Digit. Health"},{"issue":"2","key":"61342_CR2","doi-asserted-by":"publisher","first-page":"527","DOI":"10.1210\/clinem\/dgad503","volume":"109","author":"EJ Ha","year":"2024","unstructured":"Ha, E. J. et al. Artificial intelligence model assisting thyroid nodule diagnosis and management: A multicenter diagnostic study. J. Clin. Endocrinol. Metab. 109(2), 527\u2013535. https:\/\/doi.org\/10.1210\/clinem\/dgad503 (2024).","journal-title":"J. Clin. Endocrinol. Metab."},{"issue":"4","key":"61342_CR3","doi-asserted-by":"publisher","first-page":"1222","DOI":"10.3390\/jcm14041222","volume":"14","author":"E Uysal","year":"2025","unstructured":"Uysal, E., Yangoz, B., Sagan, M., Duman, I. & Alparslan, A. S. An interobserver comparison of the ultrasound lexicon classification of thyroid nodules: A single-center prospective validation study. J. Clin. Med. 14(4), 1222. https:\/\/doi.org\/10.3390\/jcm14041222 (2025).","journal-title":"J. Clin. Med."},{"key":"61342_CR4","doi-asserted-by":"publisher","DOI":"10.3389\/fonc.2022.905955","volume":"12","author":"J Yang","year":"2022","unstructured":"Yang, J. et al. Ultrasound image classification of thyroid nodules based on deep learning. Front. Oncol. 12, 905955. https:\/\/doi.org\/10.3389\/fonc.2022.905955 (2022).","journal-title":"Front. Oncol."},{"key":"61342_CR5","doi-asserted-by":"publisher","first-page":"7231","DOI":"10.1038\/s41598-023-34459-3","volume":"13","author":"M Rho","year":"2023","unstructured":"Rho, M. et al. Diagnosis of thyroid micronodules on ultrasound using a deep convolutional neural network. Sci. Rep. 13, 7231. https:\/\/doi.org\/10.1038\/s41598-023-34459-3 (2023).","journal-title":"Sci. Rep."},{"issue":"14","key":"61342_CR6","doi-asserted-by":"publisher","first-page":"2463","DOI":"10.3390\/diagnostics13142463","volume":"13","author":"OA Ajilisa","year":"2023","unstructured":"Ajilisa, O. A., Jagathy Raj, V. P. & Sabu, M. K. A deep learning framework for the characterization of thyroid nodules from ultrasound images using improved inception network and multi-level transfer learning. Diagnostics 13(14), 2463. https:\/\/doi.org\/10.3390\/diagnostics13142463 (2023).","journal-title":"Diagnostics"},{"issue":"24","key":"61342_CR7","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1177\/15353702231220664","volume":"248","author":"Y Liu","year":"2024","unstructured":"Liu, Y., Feng, Y., Qian, L., Wang, Z. & Hu, X. Deep learning diagnostic performance and visual insights in differentiating benign and malignant thyroid nodules on ultrasound images. Exp. Biol. Med. 248(24), 1\u201312. https:\/\/doi.org\/10.1177\/15353702231220664 (2024).","journal-title":"Exp. Biol. Med."},{"key":"61342_CR8","doi-asserted-by":"publisher","first-page":"1012724","DOI":"10.3389\/fonc.2022.1012724","volume":"12","author":"Y Tao","year":"2022","unstructured":"Tao, Y. et al. Deep learning for the diagnosis of suspicious thyroid nodules based on multimodal ultrasound images. Front. Oncol. 12, 1012724. https:\/\/doi.org\/10.3389\/fonc.2022.1012724 (2022).","journal-title":"Front. Oncol."},{"key":"61342_CR9","doi-asserted-by":"publisher","first-page":"4938","DOI":"10.1038\/s41598-026-35965-w","volume":"16","author":"M Saini","year":"2026","unstructured":"Saini, M. et al. Deep learning-based classification of thyroid nodules using uncertainty-aware multi-modal ultrasound imaging. Sci. Rep. 16, 4938. https:\/\/doi.org\/10.1038\/s41598-026-35965-w (2026).","journal-title":"Sci. Rep."},{"issue":"10","key":"61342_CR10","doi-asserted-by":"publisher","first-page":"255","DOI":"10.3390\/bdcc9100255","volume":"9","author":"M Savelonas","year":"2025","unstructured":"Savelonas, M. An overview of AI-guided thyroid ultrasound image segmentation and classification for nodule assessment. Big Data Cogn. Comput. 9(10), 255. https:\/\/doi.org\/10.3390\/bdcc9100255 (2025).","journal-title":"Big Data Cogn. Comput."},{"issue":"2","key":"61342_CR11","doi-asserted-by":"publisher","first-page":"27","DOI":"10.3390\/make8020027","volume":"8","author":"YA Kustiawan","year":"2026","unstructured":"Kustiawan, Y. A., Ghauth, K. I., Ghauth, S., Toong, L. Y. & Tan, S. H. Artificial intelligence techniques for thyroid cancer classification: A systematic review. Mach. Learn. Knowl. Extr. 8(2), 27 (2026).","journal-title":"Mach. Learn. Knowl. Extr."},{"key":"61342_CR12","doi-asserted-by":"publisher","DOI":"10.1016\/j.bspc.2022.103641","volume":"76","author":"W Silva","year":"2022","unstructured":"Silva, W., Teixeira, L. F. & Oliveira, A. L. Multiple instance learning for thyroid nodule classification from ultrasound sequences. Biomed. Signal Process. Control. 76, 103641. https:\/\/doi.org\/10.1016\/j.bspc.2022.103641 (2022).","journal-title":"Biomed. Signal Process. Control."},{"key":"61342_CR13","unstructured":"Ilse, M., Tomczak, J. & Welling, M. Attention-based deep multiple instance learning. In ICML 2127\u20132136 (2018)."},{"key":"61342_CR14","doi-asserted-by":"publisher","first-page":"341","DOI":"10.1186\/s12880-024-01521-z","volume":"24","author":"P Dong","year":"2024","unstructured":"Dong, P. et al. An ultrasound image segmentation method for thyroid nodules based on dual-path attention mechanism-enhanced unet++. BMC Med. Imaging 24, 341. https:\/\/doi.org\/10.1186\/s12880-024-01521-z (2024).","journal-title":"BMC Med. Imaging"},{"key":"61342_CR15","doi-asserted-by":"publisher","first-page":"1694174","DOI":"10.3389\/fmed.2026.1694174","volume":"13","author":"X Liu","year":"2026","unstructured":"Liu, X. et al. A dual-branch deep learning framework with mask-guided attention for thyroid nodule classification in ultrasound images. Front. Med. 13, 1694174. https:\/\/doi.org\/10.3389\/fmed.2026.1694174 (2026).","journal-title":"Front. Med."},{"key":"61342_CR16","doi-asserted-by":"publisher","unstructured":"Duong, V.H., Vu, H., Phan, H.D., Nguyen, D.Q., Pham, D.H., Le, Q.T., Nguyen, B.S., Do, T.D., Dinh, V.S., Nguyen, T.C., Pham, H.H. & Ngo, D.H. ThyroidXL: Advancing thyroid nodule diagnosis with an expert-labeled, pathology-validated dataset. In: Medical Image Computing and Computer Assisted Intervention \u2013 MICCAI 2025 Workshops p. 60. (Springer, 2025). https:\/\/doi.org\/10.1007\/978-3-032-05182-0_60","DOI":"10.1007\/978-3-032-05182-0_60"},{"key":"61342_CR17","doi-asserted-by":"publisher","first-page":"8073","DOI":"10.1007\/s00521-024-09505-4","volume":"36","author":"SA Balanya","year":"2024","unstructured":"Balanya, S. A., Maro\u00f1as, J. & Ramos, D. Adaptive temperature scaling for robust calibration of deep neural networks. Neural Comput. Appl. 36, 8073\u20138095. https:\/\/doi.org\/10.1007\/s00521-024-09505-4 (2024).","journal-title":"Neural Comput. Appl."},{"key":"61342_CR18","doi-asserted-by":"publisher","first-page":"2838","DOI":"10.1038\/s41591-024-03113-4","volume":"30","author":"Y Yang","year":"2024","unstructured":"Yang, Y., Zhang, H., Gichoya, J. W., Katabi, D. & Ghassemi, M. The limits of fair medical imaging AI in real-world generalization. Nat. Med. 30, 2838\u20132848. https:\/\/doi.org\/10.1038\/s41591-024-03113-4 (2024).","journal-title":"Nat. Med."},{"key":"61342_CR19","doi-asserted-by":"publisher","first-page":"203","DOI":"10.1038\/s41746-024-01196-4","volume":"7","author":"D Schwabe","year":"2024","unstructured":"Schwabe, D., Becker, K., Seyferth, M., Kla\u00df, A. & Schaeffter, T. The metric-framework for assessing data quality for trustworthy ai in medicine: A systematic review. npj Digit. Med. 7, 203. https:\/\/doi.org\/10.1038\/s41746-024-01196-4 (2024).","journal-title":"npj Digit. Med."},{"key":"61342_CR20","doi-asserted-by":"publisher","first-page":"544","DOI":"10.1186\/s12885-025-13917-3","volume":"25","author":"Y Xu","year":"2025","unstructured":"Xu, Y. et al. Thyroid nodule classification in ultrasound imaging using deep transfer learning. BMC Cancer 25, 544. https:\/\/doi.org\/10.1186\/s12885-025-13917-3 (2025).","journal-title":"BMC Cancer"},{"key":"61342_CR21","doi-asserted-by":"publisher","first-page":"40364","DOI":"10.1038\/s41598-025-25780-0","volume":"15","author":"A Rashed","year":"2025","unstructured":"Rashed, A., Medhat, T. & Elgarayhi, A. Enhancing automatic diagnosis of thyroid nodules from ultrasound scans leveraging deep learning models. Sci. Rep. 15, 40364. https:\/\/doi.org\/10.1038\/s41598-025-25780-0 (2025).","journal-title":"Sci. Rep."},{"key":"61342_CR22","doi-asserted-by":"publisher","DOI":"10.1016\/j.compbiomed.2023.107144","volume":"163","author":"W Ming","year":"2023","unstructured":"Ming, W. et al. Vision transformer-based thyroid ultrasound nodule classification with multi-scale feature fusion. Comput. Biol. Med. 163, 107144. https:\/\/doi.org\/10.1016\/j.compbiomed.2023.107144 (2023).","journal-title":"Comput. Biol. Med."},{"key":"61342_CR23","doi-asserted-by":"publisher","first-page":"32118","DOI":"10.1038\/s41598-024-83888-1","volume":"14","author":"X Yang","year":"2024","unstructured":"Yang, X. et al. Identification of lesion location and discrimination between benign and malignant findings in thyroid ultrasound imaging. Sci. Rep. 14, 32118. https:\/\/doi.org\/10.1038\/s41598-024-83888-1 (2024).","journal-title":"Sci. Rep."},{"key":"61342_CR24","doi-asserted-by":"publisher","first-page":"27207","DOI":"10.1038\/s41598-025-12602-6","volume":"15","author":"T Banerjee","year":"2025","unstructured":"Banerjee, T. et al. A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation. Sci. Rep. 15, 27207. https:\/\/doi.org\/10.1038\/s41598-025-12602-6 (2025).","journal-title":"Sci. Rep."},{"key":"61342_CR25","unstructured":"Guo, C., Pleiss, G., Sun, Y. & Weinberger, K. On calibration of modern neural networks. In: ICML pp. 1321\u20131330 (2017)."},{"issue":"9","key":"61342_CR26","doi-asserted-by":"publisher","DOI":"10.1016\/j.patter.2023.100804","volume":"4","author":"S Kapoor","year":"2023","unstructured":"Kapoor, S. & Narayanan, A. Leakage and the reproducibility crisis in machine-learning-based science. Patterns 4(9), 100804. https:\/\/doi.org\/10.1016\/j.patter.2023.100804 (2023).","journal-title":"Patterns"},{"key":"61342_CR27","doi-asserted-by":"publisher","first-page":"555","DOI":"10.1038\/s41551-020-00682-w","volume":"5","author":"MY Lu","year":"2021","unstructured":"Lu, M. Y. Data-efficient and weakly supervised computational pathology on whole-slide images. Nat. Biomed. Eng. 5, 555\u2013570. https:\/\/doi.org\/10.1038\/s41551-020-00682-w (2021).","journal-title":"Nat. Biomed. Eng."},{"key":"61342_CR28","unstructured":"Shao, Z., et al. Transmil: Transformer based correlated multiple instance learning. In: NeurIPS (2021)"},{"key":"61342_CR29","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-025-05757-4","author":"H Zhang","year":"2025","unstructured":"Zhang, H., Liu, Q., Han, X., Niu, L. & Sun, W. TN5000: An ultrasound image dataset for thyroid nodule detection and classification. Sci. Data https:\/\/doi.org\/10.1038\/s41597-025-05757-4 (2025).","journal-title":"Sci. Data"},{"issue":"3","key":"61342_CR30","doi-asserted-by":"publisher","first-page":"837","DOI":"10.2307\/2531595","volume":"44","author":"ER DeLong","year":"1988","unstructured":"DeLong, E. R., DeLong, D. M. & Clarke-Pearson, D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 44(3), 837\u2013845. https:\/\/doi.org\/10.2307\/2531595 (1988).","journal-title":"Biometrics"},{"key":"61342_CR31","doi-asserted-by":"publisher","first-page":"610","DOI":"10.1038\/s42256-021-00338-7","volume":"3","author":"AJ DeGrave","year":"2021","unstructured":"DeGrave, A. J., Janizek, J. D. & Lee, S.-I. AI for radiographic COVID-19 detection selects shortcuts over signal. Nat. Mach. Intell. 3, 610\u2013619. https:\/\/doi.org\/10.1038\/s42256-021-00338-7 (2021).","journal-title":"Nat. Mach. Intell."},{"key":"61342_CR32","doi-asserted-by":"publisher","first-page":"1808","DOI":"10.1038\/s41467-024-46000-9","volume":"15","author":"D Kiyasseh","year":"2024","unstructured":"Kiyasseh, D., Cohen, A. & Jiang, C. A framework for evaluating clinical artificial intelligence systems without ground-truth annotations. Nat. Commun. 15, 1808. https:\/\/doi.org\/10.1038\/s41467-024-46000-9 (2024).","journal-title":"Nat. Commun."},{"key":"61342_CR33","doi-asserted-by":"publisher","first-page":"29","DOI":"10.1007\/s44163-024-00119-2","volume":"4","author":"A Gerdes","year":"2024","unstructured":"Gerdes, A. The role of explainability in AI-supported medical decision-making. Discov. Artif. Intell. 4, 29. https:\/\/doi.org\/10.1007\/s44163-024-00119-2 (2024).","journal-title":"Discov. Artif. Intell."},{"key":"61342_CR34","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1007\/s44163-024-00114-7","volume":"4","author":"M Frasca","year":"2024","unstructured":"Frasca, M., La Torre, D., Pravettoni, G. & Cutica, I. Explainable and interpretable artificial intelligence in medicine: A systematic bibliometric review. Discov. Artif. Intell. 4, 15. https:\/\/doi.org\/10.1007\/s44163-024-00114-7 (2024).","journal-title":"Discov. Artif. Intell."},{"key":"61342_CR35","doi-asserted-by":"crossref","unstructured":"Wang, H., Wang, Z., Du, M., Yang, F., Zhang, Z., Ding, S., Mardziel, P. & Hu, X. Score-CAM: Score-weighted visual explanations for convolutional neural networks. In: Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops pp. 24\u201325 (2020)","DOI":"10.1109\/CVPRW50498.2020.00020"}],"container-title":["Scientific Reports"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61342-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61342-8","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61342-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T14:04:54Z","timestamp":1784210694000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61342-8"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,16]]},"references-count":35,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["61342"],"URL":"https:\/\/doi.org\/10.1038\/s41598-026-61342-8","relation":{},"ISSN":["2045-2322"],"issn-type":[{"value":"2045-2322","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,16]]},"assertion":[{"value":"22 May 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"3 July 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"16 July 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"The authors declare that they have no competing interests.","order":1,"name":"Ethics","label":"Competing interests","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"Not applicable. This study used the publicly available ThyroidXL dataset\n                      16\n                      , which was collected under ethical approval obtained by the original authors from the Research Ethics Committee of the Vietnam National Hospital of Endocrinology. All patients provided written informed consent for research use, as described in the original dataset publication.","order":2,"name":"Ethics","label":"Ethical approval","group":{"name":"EthicsHeading","label":"Declarations"}}],"article-number":"22408"}}