{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,9]],"date-time":"2026-06-09T04:59:33Z","timestamp":1780981173850,"version":"3.54.1"},"reference-count":34,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,6,9]],"date-time":"2026-06-09T00:00:00Z","timestamp":1780963200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Med."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Breast ultrasound diagnosis is significantly constrained by operator subjectivity. While Deep Learning shows promise, existing models often neglect the structured morphological semantics essential for radiological reasoning.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We propose Cross-Attention Guided Network (CGA-Net), a multi-modal framework that fuses visual data with objective clinical descriptors via a cross-attention mechanism. Specifically, clinical features\u2014such as shape and margin\u2014act as semantic queries to dynamically highlight pathological regions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Validated on 252 patients using a rigorous Out-Of-Fold (OOF) prediction strategy to prevent data leakage, the CGA-Net trained from scratch demonstrated the most balanced clinical utility, yielding a robust OOF AUC of 0.905 with the highest overall accuracy (0.857) and an optimally balanced specificity (0.831), while maintaining excellent sensitivity (0.898). Furthermore, a pre-trained version of CGA-Net achieved a peak overall ranking AUC of 0.915. Both multi-modal configurations substantially outperformed the robust clinical-only baseline (0.890) and the image-only baseline (0.795).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>This suggests that while transfer learning aids general feature extraction, strong cross-modal semantic guidance alone is highly effective at reducing false positive diagnoses and optimizing practical clinical thresholds. Attention map visualization confirmed that the model aligns closely with expert focus on tumor periphery. CGA-Net offers a robust, interpretable, and data-efficient \u201csecond opinion\u201d tool to reduce diagnostic variability.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fmed.2026.1802570","type":"journal-article","created":{"date-parts":[[2026,6,9]],"date-time":"2026-06-09T04:31:31Z","timestamp":1780979491000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Cross-attention guided multi-modal network for breast ultrasound diagnosis incorporating objective clinical semantics"],"prefix":"10.3389","volume":"13","author":[{"given":"Qin","family":"Sun","sequence":"first","affiliation":[{"name":"Department of Ultrasound Diagnostics, The First People\u2019s Hospital of Taicang, Suzhou","place":["Jiangsu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaoman","family":"Wu","sequence":"additional","affiliation":[{"name":"Department of Ultrasound Diagnostics, The First People\u2019s Hospital of Taicang, Suzhou","place":["Jiangsu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ju","family":"Chen","sequence":"additional","affiliation":[{"name":"Department of Ultrasound Diagnostics, The First People\u2019s Hospital of Taicang, Suzhou","place":["Jiangsu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yuhang","family":"Zhang","sequence":"additional","affiliation":[{"name":"Department of Ultrasound Diagnostics, The First People\u2019s Hospital of Taicang, Suzhou","place":["Jiangsu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chao","family":"Zhou","sequence":"additional","affiliation":[{"name":"Department of Ultrasound Diagnostics, The First People\u2019s Hospital of Taicang, Suzhou","place":["Jiangsu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zheng","family":"Zhu","sequence":"additional","affiliation":[{"name":"Department of Ultrasound Diagnostics, The First People\u2019s Hospital of Taicang, Suzhou","place":["Jiangsu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,6,9]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"489","DOI":"10.1016\/S1470-2045(14)70029-4","article-title":"Challenges to effective cancer control in China, India, and Russia.","volume":"15","author":"Goss","year":"2014","journal-title":"Lancet Oncol."},{"key":"B2","doi-asserted-by":"publisher","first-page":"37","DOI":"10.1016\/j.ultrasmedbio.2017.09.012","article-title":"Ultrasound imaging technologies for breast cancer detection and management: a review.","volume":"44","author":"Guo","year":"2018","journal-title":"Ultrasound Med Biol"},{"key":"B3","doi-asserted-by":"publisher","first-page":"137","DOI":"10.1186\/s13058-024-01895-6","article-title":"Deep learning applications in breast cancer histopathological imaging: diagnosis, treatment, and prognosis.","volume":"26","author":"Jiang","year":"2024","journal-title":"Breast Cancer Res"},{"key":"B4","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1186\/s42492-024-00155-w","article-title":"Multi-task approach based on combined CNN-transformer for efficient segmentation and classification of breast tumors in ultrasound images.","volume":"7","author":"Tagnamas","year":"2024","journal-title":"Vis Comput Ind Biomed Art"},{"key":"B5","doi-asserted-by":"publisher","first-page":"2026","DOI":"10.1088\/1361-6560\/ae3658","article-title":"Radiological and biological dictionary of radiomics features: addressing understandable AI issues in personalized breast cancer; dictionary version BM1.0.","volume":"71","author":"Gorji","year":"2026","journal-title":"Phys Med Biol"},{"key":"B6","doi-asserted-by":"publisher","first-page":"711","DOI":"10.1016\/j.csbj.2024.11.012","article-title":"Bridging human and machine intelligence: reverse-engineering radiologist intentions for clinical trust and adoption.","volume":"24","author":"Awasthi","year":"2024","journal-title":"Comput Struct Biotechnol J"},{"key":"B7","doi-asserted-by":"publisher","first-page":"6816","DOI":"10.3390\/s22186816","article-title":"Review of multi-modal learning from the text-guided visual processing viewpoint.","volume":"22","author":"Ullah","year":"2022","journal-title":"Sensors (Basel)"},{"key":"B8","doi-asserted-by":"publisher","first-page":"11578","DOI":"10.21037\/qims-2024-2903","article-title":"Deep learning in multi-modal breast cancer data fusion: a literature review.","volume":"15","author":"Li","year":"2025","journal-title":"Quant Imaging Med Surg"},{"key":"B9","doi-asserted-by":"publisher","first-page":"108634","DOI":"10.1016\/j.cmpb.2025.108634","article-title":"Why does my medical AI look at pictures of birds? Exploring the efficacy of transfer learning across domain boundaries.","volume":"261","author":"Jonske","year":"2025","journal-title":"Comput Methods Programs Biomed"},{"key":"B10","article-title":"Transfusion: understanding transfer learning for medical imaging.","volume-title":"Advances in Neural Information Processing Systems.","author":"Raghu","year":"2019"},{"key":"B11","volume-title":"ImageNet-Trained CNNs are Biased Towards Texture; Increasing Shape Bias Improves Accuracy and Robustness.","author":"Geirhos","year":"2018"},{"key":"B12","doi-asserted-by":"publisher","first-page":"148","DOI":"10.1038\/s41597-024-02984-z","article-title":"Curated benchmark dataset for ultrasound based breast lesion analysis.","volume":"11","author":"Paw\u0142owska","year":"2024","journal-title":"Sci Data"},{"key":"B13","volume-title":"BUS_UC.","author":"Iqbal","year":"2023"},{"key":"B14","doi-asserted-by":"publisher","first-page":"107292","DOI":"10.1016\/j.engappai.2023.107292","article-title":"Memory-efficient transformer network with feature fusion for breast tumor segmentation and classification task.","volume":"127","author":"Iqbal","year":"2023","journal-title":"Eng Appl Artif Intel."},{"key":"B15","doi-asserted-by":"publisher","first-page":"104863","DOI":"10.1016\/j.dib.2019.104863","article-title":"Dataset of breast ultrasound images.","volume":"28","author":"Al-Dhabyani","year":"2019","journal-title":"Data Brief"},{"key":"B16","doi-asserted-by":"crossref","first-page":"248","DOI":"10.1109\/CVPR.2009.5206848","article-title":"ImageNet: a large-scale hierarchical image database.","volume-title":"2009 IEEE Conference on Computer Vision and Pattern Recognition.","author":"Deng","year":"2009"},{"key":"B17","volume-title":"Deep Learning Framework for Early Detection of Pancreatic Cancer Using Multi-Modal Medical Imaging Analysis.","author":"Slobodzian","year":"2025"},{"key":"B18","doi-asserted-by":"publisher","first-page":"136","DOI":"10.1038\/s41746-020-00341-z","article-title":"Fusion of medical imaging and electronic health records using deep learning: a systematic review and implementation guidelines.","volume":"3","author":"Huang","year":"2020","journal-title":"NPJ Digit Med"},{"key":"B19","volume-title":"Attention Is All You Need.","author":"Vaswani","year":"2023"},{"key":"B20","volume-title":"PyTorch: An Imperative Style, High-Performance Deep Learning Library.","author":"Paszke","year":"2019"},{"key":"B21","volume-title":"On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima.","author":"Keskar","year":"2017"},{"key":"B22","doi-asserted-by":"publisher","first-page":"837","DOI":"10.2307\/2531595","article-title":"Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach.","volume":"44","author":"DeLong","year":"1988","journal-title":"Biometrics."},{"key":"B23","doi-asserted-by":"publisher","first-page":"622","DOI":"10.3390\/diagnostics16040622","article-title":"An interpretable ensemble transformer framework for breast cancer detection in ultrasound images.","volume":"16","author":"Al-Tam","year":"2026","journal-title":"Diagnostics (Basel)"},{"key":"B24","doi-asserted-by":"publisher","first-page":"18","DOI":"10.1055\/s-0041-1729124","article-title":"Hyperechoic lesions on breast ultrasound: all things bright and beautiful?","volume":"31","author":"Ramani","year":"2021","journal-title":"Indian J Radiol Imaging"},{"key":"B25","doi-asserted-by":"publisher","first-page":"101976","DOI":"10.1016\/j.tranon.2024.101976","article-title":"Breast imaging reporting and data system for sonography: positive and negative predictive values of sonographic features in Kumasi, Ghana.","volume":"45","author":"Badu-Peprah","year":"2024","journal-title":"Transl Oncol"},{"key":"B26","doi-asserted-by":"publisher","first-page":"349","DOI":"10.1038\/s41746-024-01328-w","article-title":"Impact of human and artificial intelligence collaboration on workload reduction in medical image interpretation.","volume":"7","author":"Chen","year":"2024","journal-title":"NPJ Digit Med"},{"key":"B27","doi-asserted-by":"publisher","first-page":"2971","DOI":"10.3390\/biomedicines10112971","article-title":"Hybrid workflow of residual convolutional transformer encoder for breast cancer classification using digital X-ray mammograms.","volume":"10","author":"Al-Tam","year":"2022","journal-title":"Biomedicines"},{"key":"B28","doi-asserted-by":"publisher","first-page":"89","DOI":"10.3390\/diagnostics13010089","article-title":"ETECADx: ensemble self-attention transformer encoder for breast cancer diagnosis using full-field digital X-ray breast images.","volume":"13","author":"Al-Hejri","year":"2022","journal-title":"Diagnostics (Basel)"},{"key":"B29","doi-asserted-by":"publisher","first-page":"731","DOI":"10.1016\/j.bbe.2024.08.007","article-title":"Multimodal breast cancer hybrid explainable computer-aided diagnosis using medical mammograms and ultrasound images.","volume":"44","author":"Al-Tam","year":"2024","journal-title":"Biocybern Biomed Eng."},{"key":"B30","doi-asserted-by":"publisher","first-page":"29224","DOI":"10.1038\/s41598-024-79838-6","article-title":"The risk of shortcutting in deep learning algorithms for medical imaging research.","volume":"14","author":"Hill","year":"2024","journal-title":"Sci Rep"},{"key":"B31","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1007\/s40134-018-0266-8","article-title":"BI-RADS 3: current and future use of probably benign.","volume":"6","author":"Lee","year":"2018","journal-title":"Curr Radiol Rep"},{"key":"B32","doi-asserted-by":"publisher","first-page":"2176","DOI":"10.1038\/s41591-021-01595-0","article-title":"Underdiagnosis bias of artificial intelligence algorithms applied to chest radiographs in under-served patient populations.","volume":"27","author":"Seyyed-Kalantari","year":"2021","journal-title":"Nat Med"},{"key":"B33","doi-asserted-by":"publisher","first-page":"133","DOI":"10.1186\/s12880-024-01307-3","article-title":"Spatial and geometric learning for classification of breast tumors from multi-center ultrasound images: a hybrid learning approach.","volume":"24","author":"Ru","year":"2024","journal-title":"BMC Med Imaging"},{"key":"B34","doi-asserted-by":"publisher","first-page":"1134","DOI":"10.1038\/s41591-024-02855-5","article-title":"Adapted large language models can outperform medical experts in clinical text summarization.","volume":"30","author":"Van Veen","year":"2024","journal-title":"Nat Med"}],"container-title":["Frontiers in Medicine"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1802570\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,9]],"date-time":"2026-06-09T04:32:16Z","timestamp":1780979536000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1802570\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,9]]},"references-count":34,"alternative-id":["10.3389\/fmed.2026.1802570"],"URL":"https:\/\/doi.org\/10.3389\/fmed.2026.1802570","relation":{},"ISSN":["2296-858X"],"issn-type":[{"value":"2296-858X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,9]]},"article-number":"1802570"}}