Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

A foundation generative model for breast ultrasound image analysis

Nature Biomedical Engineering (2026) Cite this article

Subjects

Abstract

Foundation models have emerged as powerful tools for addressing various tasks in clinical settings. However, their potential development for breast ultrasound analysis remains untapped. Here we present BUSGen, the first foundation generative model designed for breast ultrasound image analysis. Pretrained on over 3.5 million breast ultrasound images, BUSGen has acquired extensive knowledge of breast structures, pathological features and clinical variations. With few-shot adaptation, BUSGen can generate repositories of realistic and informative task-specific data, facilitating the development of models for a wide range of downstream tasks. Extensive experiments highlight BUSGen’s exceptional adaptability, significantly exceeding real-data-trained foundation models in breast cancer screening, diagnosis and prognosis. In breast cancer early diagnosis, our approach outperformed all board-certified radiologists (n = 9), achieving an average sensitivity improvement of 16.5% (P < 0.0001). In addition, we characterized the scaling effect of using synthetic data. Finally, BUSGen enabled de-identified data sharing, making progress forward in secure medical data utilization.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: The schematic overview of the BUSGen pretraining and adaptation framework.
The alternative text for this image may have been generated using AI.
Fig. 2: BUSGen can generate realistic data while protecting patient privacy.
The alternative text for this image may have been generated using AI.
Fig. 3: BUSGen improves breast cancer screening tasks.
The alternative text for this image may have been generated using AI.
Fig. 4: BUSGen enhances breast ultrasound diagnosis.
The alternative text for this image may have been generated using AI.
Fig. 5: BUSGen facilitates breast cancer prognosis.
The alternative text for this image may have been generated using AI.

Similar content being viewed by others

Data availability

The BUSI dataset used in this study is publicly available at https://www.kaggle.com/datasets/aryashah2k/breast-ultrasound-images-dataset. The UDIAT dataset is publicly available at https://www2.docm.mmu.ac.uk/STAFF/m.yap/dataset.php. The BUS-BRA103 dataset is publicly available at https://zenodo.org/records/8231412. We released a repository of the BUSI dataset with scanner device augmentation, publicly available in figshare at https://doi.org/10.6084/m9.figshare.30571916.v1 (ref. 104). However, due to respective Institutional Review Board restrictions and to protect patient privacy, the BUS-3.5M and the downstream datasets used in this study cannot be made publicly available. De-identified data may be made available by the corresponding authors for research purposes upon reasonable request. Source data are provided with this paper.

Code availability

The pretraining and adaptation code for BUSGen and an online API are available upon reasonable request via email to haojunyu@pku.edu.cn. Requests will be responded to within 6 weeks. Also, we provide an online demo of BUSGen at https://aibus.bio.

References

  1. Siegel, R. L. et al. Cancer statistics, 2023. CA Cancer J. Clin. 73, 17–48 (2023).

    PubMed  Google Scholar 

  2. Chhikara, B. S. & Parang, K. Global cancer statistics 2022: the trends projection analysis. Chem. Biol. Lett. 10, 451–451 (2023).

    Google Scholar 

  3. Xia, C. et al. Cancer statistics in China and United States, 2022: profiles, trends, and determinants. Chin. Med. J. 135, 584–590 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

  4. Sickles, E. A. ACR BI-RADS atlas, breast imaging reporting and data system. J. Am. Coll. Radiol. 39 (2013).

  5. Ohuchi, N. et al. Sensitivity and specificity of mammography and adjunctive ultrasonography to screen for breast cancer in the Japan Strategic Anti-cancer Randomized Trial (J-START): a randomised controlled trial. Lancet 387, 341–348 (2016).

    Article  PubMed  Google Scholar 

  6. Berg, W. A. et al. Detection of breast cancer with addition of annual screening ultrasound or a single screening MRI to mammography in women with elevated breast cancer risk. JAMA 307, 1394–1404 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Shen, S. et al. A multi-centre randomised trial comparing ultrasound vs mammography for screening breast cancer in high-risk Chinese women. Br. J. Cancer 112, 998–1004 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Brem, R. F., Lenihan, M. J., Lieberman, J. & Torrente, J. Screening breast ultrasound: past, present, and future. Am. J. Roentgenol. 204, 234–240 (2015).

    Article  Google Scholar 

  9. Sood, R. et al. Ultrasound for breast cancer detection globally: a systematic review and meta-analysis. J. Glob. Oncol. 5, 1–17 (2019).

  10. Park, Y. et al. Pan-Asian adapted ESMO clinical practice guidelines for the management of patients with early breast cancer: a KSMO-ESMO initiative endorsed by CSCO, ISMPO, JSMO, MOS, SSO and TOS. Ann. Oncol. 31, 451–469 (2020).

    Article  CAS  PubMed  Google Scholar 

  11. Weese, J. & Lorenz, C. Four Challenges in Medical Image Analysis from an Industrial Perspective (Elsevier, 2016).

  12. González-Villà, S. et al. A review on brain structures segmentation in magnetic resonance imaging. Artif. Intell. Med. 73, 45–69 (2016).

    Article  PubMed  Google Scholar 

  13. Geirhos, R. et al. Shortcut learning in deep neural networks. Nat. Mach. Intell. 2, 665–673 (2020).

    Article  Google Scholar 

  14. Cao, K. et al. Large-scale pancreatic cancer detection via non-contrast CT and deep learning. Nat. Med. 29, 3033–3043 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Goetz, L., Seedat, N., Vandersluis, R. & Schaar, M. Generalization—a key challenge for responsible AI in patient-facing clinical applications. npj Digit. Med. 7, 126 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  16. Tiu, E. et al. Expert-level detection of pathologies from unannotated chest X-ray images via self-supervised learning. Nat. Biomed. Eng. 6, 1399–1406 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

  17. Huang, Z., Bianchi, F., Yuksekgonul, M., Montine, T. J. & Zou, J. A visual–language foundation model for pathology image analysis using medical Twitter. Nat. Med. 29, 2307–2316 (2023).

    Article  CAS  PubMed  Google Scholar 

  18. Xu, H. et al. A whole-slide foundation model for digital pathology from real-world data. Nature 630, 181–188 (2024).

  19. De Fauw, J. et al. Clinically applicable deep learning for diagnosis and referral in retinal disease. Nat. Med. 24, 1342–1350 (2018).

    Article  PubMed  Google Scholar 

  20. Zhou, Y. et al. A foundation model for generalizable disease detection from retinal images. Nature 622, 156–163 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Yap, M. H. et al. Automated breast ultrasound lesions detection using convolutional neural networks. IEEE J. Biomed. Health Inform. 22, 1218–1226 (2017).

    Article  PubMed  Google Scholar 

  22. Al-Dhabyani, W., Gomaa, M., Khaled, H. & Fahmy, A. Dataset of breast ultrasound images. Data Brief 28, 104863 (2020).

    Article  PubMed  Google Scholar 

  23. Lin, Z. et al. A new dataset and a baseline model for breast lesion detection in ultrasound videos. In International Conference on Medical Image Computing and Computer-Assisted Intervention (eds Wang, L. et al.) 614–623 (Springer, 2022).

  24. Malin, B. A., Emam, K. E. & O’Keefe, C. M. Biomedical data privacy: problems, perspectives, and recent advances. J. Am. Med. Inform. Assoc. 20, 2–6 (2013).

    Article  PubMed  Google Scholar 

  25. Shen, Y. et al. Artificial intelligence system reduces false-positive findings in the interpretation of breast ultrasound exams. Nat. Commun. 12, 5645 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Wu, T., Sultan, L. R., Tian, J., Cary, T. W. & Sehgal, C. M. Machine learning for diagnostic ultrasound of triple-negative breast cancer. Breast Cancer Res. Treat. 173, 365–373 (2019).

    Article  CAS  PubMed  Google Scholar 

  27. Ozaki, J. et al. Deep learning method with a convolutional neural network for image classification of normal and metastatic axillary lymph nodes on breast ultrasonography. Jpn J. Radiol. 40, 814–822 (2022).

    Article  PubMed  Google Scholar 

  28. Zuluaga-Gomez, J., Al Masry, Z., Benaggoune, K., Meraghni, S. & Zerhouni, N. A CNN-based methodology for breast cancer diagnosis using thermal images. Comput. Methods Biomech. Biomed. Eng. Imaging Vis. 9, 131–145 (2021).

    Article  Google Scholar 

  29. Ong Ly, C. et al. Shortcut learning in medical AI hinders generalization: method for estimating AI model generalization without external data. npj Digit. Med. 7, 124 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  30. Namli, T. et al. A scalable and transparent data pipeline for AI-enabled health data ecosystems. Front. Med. 11, 1393123 (2024).

    Article  Google Scholar 

  31. Brown, T. et al. Language models are few-shot learners. Adv. Neural Inf. Process. Syst. 33, 1877–1901 (2020).

    Google Scholar 

  32. Ouyang, L. et al. Training language models to follow instructions with human feedback. Adv. Neural Inf. Process. Syst. 35, 27730–27744 (2022).

    Google Scholar 

  33. Rombach, R., Blattmann, A., Lorenz, D., Esser, P. & Ommer, B. High-resolution image synthesis with latent diffusion models. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition 10684–10695 (IEEE, 2022).

  34. Gemini Team Google. Gemini: a family of highly capable multimodal models. Preprint at https://arxiv.org/abs/2312.11805 (2023).

  35. Bommasani, R. et al. On the opportunities and risks of foundation models. Preprint at https://arxiv.org/abs/2108.07258 (2021).

  36. Bluethgen, C. et al. A vision-language foundation model for the generation of realistic chest X-ray images. Nat. Biomed. Eng. 9, 494–506 (2025).

    Article  PubMed  Google Scholar 

  37. Wang, J. et al. Self-improving generative foundation model for synthetic medical image generation and clinical applications. Nat. Med. 31, 609–617 (2025).

  38. McKinney, S. M. et al. International evaluation of an AI system for breast cancer screening. Nature 577, 89–94 (2020).

    Article  CAS  PubMed  Google Scholar 

  39. Lee, C. H. et al. Breast cancer screening with imaging: recommendations from the society of breast imaging and the ACR on the use of mammography, breast MRI, breast ultrasound, and other technologies for the detection of clinically occult breast cancer. J. Am. Coll. Radiol. 7, 18–27 (2010).

    Article  PubMed  Google Scholar 

  40. Marmot, M. G. et al. The benefits and harms of breast cancer screening: an independent review. Br. J. Cancer 108, 2205–2240 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Qian, X. et al. Prospective assessment of breast cancer risk from multimodal multiview ultrasound images via clinically applicable deep learning. Nat. Biomed. Eng. 5, 522–532 (2021).

    Article  PubMed  Google Scholar 

  42. Pacilè, S. et al. Improving breast cancer detection accuracy of mammography with the concurrent use of an artificial intelligence tool. Radiol. Artif. Intell. 2, 190208 (2020).

    Article  Google Scholar 

  43. Dembrower, K. et al. Effect of artificial intelligence-based triaging of breast cancer screening mammograms on cancer detection and radiologist workload: a retrospective simulation study. Lancet Digit. Health 2, 468–474 (2020).

    Article  Google Scholar 

  44. Zhou, L.-Q. et al. Lymph node metastasis prediction from primary breast cancer US images using deep learning. Radiology 294, 19–28 (2020).

    Article  PubMed  Google Scholar 

  45. Tafreshi, N. K., Kumar, V., Morse, D. L. & Gatenby, R. A. Molecular and functional imaging of breast cancer. Cancer Control 17, 143–155 (2010).

    Article  PubMed  Google Scholar 

  46. Zhao, S., Zuo, W.-J., Shao, Z.-M. & Jiang, Y.-Z. Molecular subtypes and precision treatment of triple-negative breast cancer. Ann. Transl. Med. 8, 499 (2020).

  47. Tsang, J. Y. & Tse, G. M. Molecular classification of breast cancer. Adv. Anat. Pathol. 27, 27–35 (2020).

    Article  CAS  PubMed  Google Scholar 

  48. Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models. Adv. Neural Inf. Process. Syst. 33, 6840–6851 (2020).

    Google Scholar 

  49. Radford, A. et al. Learning transferable visual models from natural language supervision. In International Conference on Machine Learning 8748–8763 (PMLR, 2021).

  50. Fan, L. et al. Scaling laws of synthetic images for model training… for now. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition 7382–7392 (IEEE, 2024).

  51. Ktena, I. et al. Generative models improve fairness of medical classifiers under distribution shifts. Nat. Med. 30, 1166–1173 (2024).

  52. Dhariwal, P. & Nichol, A. Diffusion models beat GANS on image synthesis. Adv. Neural Inf. Process. Syst. 34, 8780–8794 (2021).

    Google Scholar 

  53. Ho, J. & Salimans, T. Classifier-free diffusion guidance. In NeurIPS Workshop on Deep Generative Models and Downstream Applications Poster (NeurIPS, 2021).

  54. Zhang, L., Rao, A. & Agrawala, M. Adding conditional control to text-to-image diffusion models. In Proc. IEEE/CVF International Conference on Computer Vision 3836–3847 (IEEE,2023).

  55. Hu, E. J. et al. LoRA: Low-rank adaptation of large language models. In International Conference on Learning Representations Poster (ICLR, 2022).

  56. Lu, C. et al. DPM-Solver: a fast ODE solver for diffusion probabilistic model sampling in around 10 steps. Adv. Neural Inf. Process. Syst. 35, 5775–5787 (2022).

    Google Scholar 

  57. Lu, C. et al. DPM-Solver++: fast solver for guided sampling of diffusion probabilistic models. Mach. Intell. Res. 22, 730–751 (2025).

    Article  Google Scholar 

  58. Kazdan, J. et al. CPsample: classifier protected sampling for guarding training data during diffusion. In Proc. 13th International Conference on Learning Representations (ICLR, 2025).

  59. Lin, W. et al. PMC-CLIP: contrastive language-image pre-training using biomedical documents. In Proc. International Conference on Medical Image Computing and Computer-Assisted Intervention 525–536 (Springer, 2023).

  60. Wang, Z., Wu, Z., Agarwal, D. & Sun, J. MedCLIP: contrastive learning from unpaired medical images and text. In Proc. Conference on Empirical Methods in Natural Language Processing 3876 (ACL, 2022).

  61. Eslami, S., Meinel, C. & de Melo, G. in Findings of the Association for Computational Linguistics: EACL 2023 (eds Vlachos, A. & Augenstein, I.) 1181–1193 (ACL, 2023).

  62. Zhang, S. et al. A multimodal biomedical foundation model trained from fifteen million image–text pairs. NEJM AI 2, 2400640 (2025).

    Article  Google Scholar 

  63. Chen, X., Xie, S. & He, K. An empirical study of training self-supervised vision transformers. In Proc. IEEE/CVF International Conference on Computer Vision 9640–9649 (IEEE, 2021).

  64. He, K. et al. Masked autoencoders are scalable vision learners. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition 16000–16009 (IEEE, 2022).

  65. Geman, D., Geman, S., Hallonquist, N. & Younes, L. Visual Turing test for computer vision systems. Proc. Natl Acad. Sci. USA 112, 3618–3623 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  66. Carlini, N. et al. Extracting training data from large language models. In 30th USENIX Security Symposium (USENIX Security 21) 2633–2650 (USENIX, 2021).

  67. Elmore, J. G., Armstrong, K., Lehman, C. D. & Fletcher, S. W. Screening for breast cancer. JAMA 293, 1245–1256 (2005).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  68. Myers, E. R. et al. Benefits and harms of breast cancer screening: a systematic review. JAMA 314, 1615–1634 (2015).

    Article  CAS  PubMed  Google Scholar 

  69. Yuan, W.-H., Hsu, H.-C., Chen, Y.-Y. & Wu, C.-H. Supplemental breast cancer-screening ultrasonography in women with dense breasts: a systematic review and meta-analysis. Br. J. Cancer 123, 673–688 (2020).

    Article  PubMed  PubMed Central  Google Scholar 

  70. Yap, M. H. et al. Breast ultrasound region of interest detection and lesion localisation. Artif. Intell. Med. 107, 101880 (2020).

    Article  PubMed  Google Scholar 

  71. Pinder, S. E. Ductal carcinoma in situ (DCIS): pathological features, differential diagnosis, prognostic factors and specimen evaluation. Modern Pathol. 23, 8–13 (2010).

    Article  Google Scholar 

  72. Ernster, V. L. & Barclay, J. Increases in ductal carcinoma in situ (DCIS) of the breast in relation to mammography: a dilemma. J. Natl Cancer Inst. Monogr. 1997, 151–156 (1997).

    Article  Google Scholar 

  73. Winchester, D. P., Jeske, J. M. & Goldschmidt, R. A. The diagnosis and management of ductal carcinoma in-situ of the breast. CA Cancer J. Clin. 50, 184–200 (2000).

    CAS  PubMed  Google Scholar 

  74. Watanabe, T. et al. Ultrasound image classification of ductal carcinoma in situ (DCIS) of the breast: analysis of 705 DCIS lesions. Ultrasound Med. Biol. 43, 918–925 (2017).

    Article  PubMed  Google Scholar 

  75. Kim, S. H. et al. Correlation of ultrasound findings with histology, tumor grade, and biological markers in breast cancer. Acta Oncol. 47, 1531–1538 (2008).

    Article  CAS  PubMed  Google Scholar 

  76. Ahsan, M. M., Mahmud, M. P., Saha, P. K., Gupta, K. D. & Siddique, Z. Effect of data scaling methods on machine learning algorithms and model performance. Technologies 9, 52 (2021).

    Article  Google Scholar 

  77. Frid-Adar, M. et al. GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classification. Neurocomputing 321, 321–331 (2018).

    Article  Google Scholar 

  78. Sagers, L. W. et al. Augmenting medical image classifiers with synthetic data from latent diffusion models. Preprint at https://arxiv.org/abs/2308.12453 (2023).

  79. Kerlikowske, K., Smith-Bindman, R., Ljung, B.-M. & Grady, D. Evaluation of abnormal mammography results and palpable breast abnormalities. Ann. Intern. Med. 139, 274–284 (2003).

    Article  PubMed  Google Scholar 

  80. Weigel, M. T. & Dowsett, M. Current and emerging biomarkers in breast cancer: prognosis and prediction. Endocr. Relat. Cancer 17, 245–262 (2010).

    Article  Google Scholar 

  81. Baxevanis, C. N., Fortis, S. P. & Perez, S. A. in Seminars in Cancer Biology Vol. 72 76–89 (Elsevier, 2021).

  82. Liedtke, C. et al. Response to neoadjuvant therapy and long-term survival in patients with triple-negative breast cancer. J. Clin. Oncol. 26, 1275–1281 (2008).

    Article  PubMed  Google Scholar 

  83. Kennecke, H. et al. Metastatic behavior of breast cancer subtypes. J. Clin. Oncol. 28, 3271–3277 (2010).

    Article  PubMed  Google Scholar 

  84. Rao, R., Euhus, D., Mayo, H. G. & Balch, C. Axillary node interventions in breast cancer: a systematic review. JAMA 310, 1385–1394 (2013).

    Article  CAS  PubMed  Google Scholar 

  85. Cianfrocca, M. & Goldstein, L. J. Prognostic and predictive factors in early-stage breast cancer. Oncologist 9, 606–616 (2004).

    Article  PubMed  Google Scholar 

  86. Selvaraju, R. R. et al. Grad-CAM: visual explanations from deep networks via gradient-based localization. In Proc. IEEE International Conference on Computer Vision 618–626 (IEEE, 2017).

  87. Gatys, L. A., Ecker, A. S. & Bethge, M. Image style transfer using convolutional neural networks. In Proc. IEEE Conference on Computer Vision and Pattern Recognition 2414–2423 (IEEE, 2016).

  88. Zhu, J.-Y., Park, T., Isola, P. & Efros, A. A. Unpaired image-to-image translation using cycle-consistent adversarial networks. In Proc. IEEE International Conference on Computer Vision 2223–2232 (IEEE, 2017).

  89. He, K., Gkioxari, G., Dollár, P. & Girshick, R. Mask R-CNN. In Proc. IEEE International Conference on Computer Vision 2961–2969 (IEEE, 2017).

  90. 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, 837–845 (1988).

  91. Sun, X. & Xu, W. Fast implementation of DeLong’s algorithm for comparing the areas under correlated receiver operating characteristic curves. IEEE Signal Process. Lett. 21, 1389–1393 (2014).

    Article  Google Scholar 

  92. Benjamini, Y., Drai, D., Elmer, G., Kafkafi, N. & Golani, I. Controlling the false discovery rate in behavior genetics research. Behav.Brain Res. 125, 279–284 (2001).

    Article  CAS  PubMed  Google Scholar 

  93. Glickman, M. E., Rao, S. R. & Schultz, M. R. False discovery rate control is a recommended alternative to Bonferroni-type adjustments in health studies. J. Clin. Epidemiol. 67, 850–857 (2014).

    Article  PubMed  Google Scholar 

  94. Hsu, W. et al. External validation of an ensemble model for automated mammography interpretation by artificial intelligence. JAMA Netw. Open 5, 2242343 (2022).

    Article  Google Scholar 

  95. Park, S. H. et al. Methods for clinical evaluation of artificial intelligence algorithms for medical diagnosis. Radiology 306, 20–31 (2023).

    Article  PubMed  Google Scholar 

  96. Ahn, J. S. et al. Association of artificial intelligence-aided chest radiograph interpretation with reader performance and efficiency. JAMA Netw. Open 5, 2229289 (2022).

    Article  Google Scholar 

  97. Twilt, J. J. et al. AI-assisted vs unassisted identification of prostate cancer in magnetic resonance images. JAMA Netw. Open 8, 2515672 (2025).

    Article  Google Scholar 

  98. Gommers, J. J. et al. Influence of AI decision support on radiologists’ performance and visual search in screening mammography. Radiology 316, 243688 (2025).

    Article  Google Scholar 

  99. Hunter, D. J. & Holmes, C. Where medical statistics meets artificial intelligence. N. Engl. J. Med. 389, 1211–1219 (2023).

    Article  PubMed  Google Scholar 

  100. Hwang, Y.-T. & Su, N.-C. Sample size determination for comparing accuracies between two diagnostic tests under a paired design. Biom. J. 64, 771–804 (2022).

    Article  PubMed  Google Scholar 

  101. Riesthuis, P., Otgaar, H. & Bücken, C. Ready to ROC? A tutorial on simulation-based power analyses for null hypothesis significance, minimum-effect, and equivalence testing for ROC curve analyses. Behav. Res. Methods 57, 1–20 (2025).

    Article  Google Scholar 

  102. Jung, S.-H. A Dunnett-type test and its sample size calculation for comparing K ROC curves with a control. Diagnostics 14, 1813 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  103. Gómez-Flores, W., Gregorio-Calas, M. J. & de Albuquerque Pereira, W. C. BUS-BRA: a breast ultrasound dataset for assessing computer-aided diagnosis systems. Med. Phys. 51, 3110–3123 (2024).

    Article  PubMed  Google Scholar 

  104. Li, Y. BUSI dataset with scanner device augmentation. figshare https://doi.org/10.6084/m9.figshare.30571916.v1 (2025).

Download references

Acknowledgements

We thank R. Li and J. Fan from Peking University for helpful suggestion and discussion. D.H. was supported by a National Science and Technology Major Project (2022ZD0160300) and the National Science Foundation of China (NSFC62376007). L.W. was supported by the National Science Foundation of China (NSFC92470123, NSFC62276005).

Author information

Author notes

  1. These authors contributed equally: Haojun Yu, Youcheng Li, Nan Zhang, Zihan Niu.

Authors and Affiliations

  1. State Key Laboratory of General Artificial Intelligence, Peking University, Beijing, China

    Haojun Yu, Youcheng Li, Di Dai, Di He & Liwei Wang

  2. Key laboratory of Carcinogenesis and Translational Research (Ministry of Education), Breast Center of department, Peking University Cancer Hospital and Institute, Beijing, China

    Nan Zhang, Wangyan Qin & Ling Huo

  3. Peking Union Medical College Hospital, Beijing, China

    Zihan Niu, Yanwen Luo, Mengyuan Zhou, Jia Tao & Qingli Zhu

  4. Department of Ultrasound, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China

    Xuantong Gong, Jie Han & Yong Wang

  5. Department of Computer Sciences, Stanford University, Stanford, CA, USA

    Haotian Ye & James Zou

  6. Department of Biomedical Data Science, Stanford University, Stanford, CA, USA

    Siyu He & James Zou

  7. Center for Data Science, Peking University, Beijing, China

    Quanlin Wu

  8. Yizhun Medical AI, Beijing, China

    Ziwei Zhao & Dong Wang

  9. Nanchang People’s Hospital, Nanchang, China

    Binghui Tang

  10. Department of Ultrasound, The First Affiliated Hospital of China Medical University, Shenyang, China

    Yong Wang

  11. Center for Machine Learning Research, Peking University, Beijing, China

    Liwei Wang

Authors
  1. Haojun Yu
    View author publications

    Search author on:PubMed Google Scholar

  2. Youcheng Li
    View author publications

    Search author on:PubMed Google Scholar

  3. Nan Zhang
    View author publications

    Search author on:PubMed Google Scholar

  4. Zihan Niu
    View author publications

    Search author on:PubMed Google Scholar

  5. Xuantong Gong
    View author publications

    Search author on:PubMed Google Scholar

  6. Yanwen Luo
    View author publications

    Search author on:PubMed Google Scholar

  7. Haotian Ye
    View author publications

    Search author on:PubMed Google Scholar

  8. Siyu He
    View author publications

    Search author on:PubMed Google Scholar

  9. Quanlin Wu
    View author publications

    Search author on:PubMed Google Scholar

  10. Wangyan Qin
    View author publications

    Search author on:PubMed Google Scholar

  11. Mengyuan Zhou
    View author publications

    Search author on:PubMed Google Scholar

  12. Jie Han
    View author publications

    Search author on:PubMed Google Scholar

  13. Jia Tao
    View author publications

    Search author on:PubMed Google Scholar

  14. Ziwei Zhao
    View author publications

    Search author on:PubMed Google Scholar

  15. Di Dai
    View author publications

    Search author on:PubMed Google Scholar

  16. Di He
    View author publications

    Search author on:PubMed Google Scholar

  17. Dong Wang
    View author publications

    Search author on:PubMed Google Scholar

  18. Binghui Tang
    View author publications

    Search author on:PubMed Google Scholar

  19. Ling Huo
    View author publications

    Search author on:PubMed Google Scholar

  20. James Zou
    View author publications

    Search author on:PubMed Google Scholar

  21. Qingli Zhu
    View author publications

    Search author on:PubMed Google Scholar

  22. Yong Wang
    View author publications

    Search author on:PubMed Google Scholar

  23. Liwei Wang
    View author publications

    Search author on:PubMed Google Scholar

Contributions

H. Yu and L.W. conceived and designed the study. Z.N., B.T., Y. Luo and X.G. carried out data acquisition. Y. Li, H. Yu, Y. Luo, Z.N. and Q.W. carried out data processing and annotation. H. Yu developed the AI models. Y. Li carried out generated data cleaning. Y. Li developed the platform for reader study. N.Z., Z.N., W.Q., J.T., M.Z., X.G., J.H., L.H. and Y.W. participated in the reader study. H. Yu, H. Ye, S.H., D.H., Y. Li, N.Z., Z.N., D.W., Z.Z., Q.W., D.D., Q.Z., J.Z. and L.W. wrote and revised the paper.

Corresponding authors

Correspondence to Di He, Dong Wang, Binghui Tang, Ling Huo, James Zou, Qingli Zhu, Yong Wang or Liwei Wang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Biomedical Engineering thanks Fajin Dong and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Overview of BUSGen pretraining and fine-tuning.

a, The denoising and noising process of diffusion model, where xT represents the noisy image and x0 is the original image. The model learns to denoise xt to obtain xt−1 iteratively. b, The architecture of BUSGen demonstrates the U-Net structure with incorporated condition embeddings and time embeddings at each layer. The structure includes multiple encoder blocks, a middle block, and decoder blocks, where each encoder block consists of ResBlocks and downsampling, the middle block contains ResBlocks and an AttnBlock, and each decoder block comprises ResBlocks and upsampling. c, Encoder, Middle, and Decoder Blocks. d, ResBlock and AttnBlock. e, Illustration of the LoRA fine-tuning principle, where pretrained weights W are factorized into low-rank matrices A and B, enabling efficient adaptation of the model.

Extended Data Fig. 2 Comparison of BUS-DM to state-of-the-art models.

The baselines (MedCLIP, PubMedCLIP and BiomedCLIP) were pretrained on large-scale public available image-text pairs.

Extended Data Fig. 3 Performance comparison with self-supervised learning baselines.

BUS-DM outperforms Baseline-CLIP, ViT-MoCo, and ViT-MAE across benign-malignant classification (a-d), molecular subtype classification (e), and ALN status classification (f) tasks.

Extended Data Fig. 4 DABIS score of real data and generated data.

DABIS score comparison. Real data shows higher shortcut learning (AUC 0.600) compared to BUSGen-generated data (AUC 0.493), indicating that generated data reduces data acquisition-induced bias.

Source data

Extended Data Fig. 5 Saliency maps of the ALN metastasis prediction task.

Saliency maps for ALN metastasis prediction. Heat maps (bottom row) highlight peritumoral regions most influential for predicting axillary lymph node metastasis from ultrasound images (top row).

Extended Data Table 1 Patient demographics and lesion characteristics of BUS-3.5M
Full size table
Extended Data Table 2 Benign and malignant subtypes of lesions in the BUS-3.5M
Full size table
Extended Data Table 3 Downstream datasets
Full size table
Extended Data Table 4 Reader Experience
Full size table
Extended Data Table 5 Detailed results of reader study
Full size table

Supplementary information

Supplementary Information (download PDF )

Supplementary Sections 1–9, Figs. 1–14, Tables 1–18 and References.

Reporting Summary (download PDF )

Peer Review File (download PDF )

Source data

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Yu, H., Li, Y., Zhang, N. et al. A foundation generative model for breast ultrasound image analysis. Nat. Biomed. Eng (2026). https://doi.org/10.1038/s41551-026-01639-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • DOI: https://doi.org/10.1038/s41551-026-01639-1

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing