Applications of Generative Artificial Intelligence in Medical Imaging: A Review of Opportunities and Risks
Main Article Content
Keywords
generative artificial intelligence, medical imaging, large language models, medical hallucination, smart, healthcare
Abstract
Generative Artificial Intelligence (GAI), with its core capabilities in image synthesis, data augmentation, and cross-modal transformation, is profoundly reshaping the entire workflow of medical imaging, including acquisition, diagnosis, treatment planning, and ed ucation and training. This paper systematically reviews the technical pathways and clinical application progress of generative AI in the field of medical imaging. It focuses on analyzing its core value in improving imaging quality, alleviating data scarcity, assisting precision diagnosis and treatment, and empowering primary healthcare. At the same time, it deeply explores potential risks such as “medical hallucination” misleading, black -box effects, data bias, and lagging compliance regulation. On this bas is, the paper proposes development strategies emphasizing compliance, standardization, and strong interpretability, aiming to provide theoretical references for the safe implementation and large-scale promotion of generative AI in the field of intelligent medical imaging.
References
- [1] Zhang, R. R., Pu, L. H., Zhang, W., Qi, X. Y., & Yang, F. (2019). Current status and prospects of artificial intelligence applications in medical imaging education. Modern Preventive Medicine , 46(24), 4527-4609.
- [2] Tang, B. ( 2024). Generative artificial intelligence empowering new quality productive forces: A dual deconstruction based on Marxist productivity theory and new development concepts. Academic Forum, (12), 68-81.
- [3] Peng, C. (2025). Research on generative artificial intelligence in visual art creation [Master's thesis, Donghua University]. https://doi.org/10.27012/d.cnki.gdhuu.2025.000326.
- [4] Song, Y., Shen, L., Xing, L., & Ermon, S. (2022). Solving inverse problems in medical imaging with score-based generative models. IEEE Transactions on Medical Imaging, 41(4), 818-828.
- [5] Zhu, B., Liu, J. Z., Cauley, S. F., Rosen, B. R., & Rosen, M. S. (2018). Image reconstruction by domain- transform manifold learning. Nature, 555(7694): 60-65.
- [6] Sandfort, V., Yan, K., Pickhardt, P. J., et al. (2019). Data augmentation using generative adversarial networks (CycleGAN) to improve generalizability in CT segmentation tasks. Scientific Reports , 9(1), Article 16884.
- [7] Yue, Y. Y. (2022). Research on medical image enhancement algorit hms based on generative adversarial networks [Master's thesis, Nanjing University of Posts and Telecommunications].
- [8] Rieke, N., Hancox, J., Li, W., et al. (2020). The future of digital health with federated learning. NPJ Digital Medicine, 3(1), 119.
- [9] Zhang, H. Y., Wang, T. B., Li, M. Z., Zhao, Z., Pu, S. L., & Wu, F. (2022). A review of visual-language multimodal pre -training. Journal of Image and Graphics , 27(9), 2652 -2682. https://doi.org/10.11834/jig.220173.
- [10] Tu, T., Azizi, S., Driess, D., et al. (2024). Towards generalist biomedical AI. NEJM AI, 1(3), Article AIoa2300138.
- [11] Maspero, M., Savenije, M. H. F., Dinkla, A. M., et al. (2018). Dose evaluation of fast synthetic -CT generation using a generative adversarial network for general pelvis MR -only radiotherapy. Physics in Medicine & Biology, 63(18), 18500.
- [12] Che, L., Zhang, Z. Q., Zhou, J. J., & Li, L. (2024). Current research status and development trends of generative artificial intelligence. Science and Technology Review , 42(12), 35 -43. https://doi.org/10.3981/j.issn.1000-7857.2023.03.00366
- [13] Cohen, J. P., Luck, M., & Honari, S. (2018). Distribution matching losses can hallucinate features in medical image translation. In International Conference on Medical Image Computing and Computer - Assisted Intervention (MICCAI) (pp. 529-536). Springer.
- [14] Seyyed-Kalantari, L., Zhang, H., McDermott, M. B. A., et al. (2021). Underdiagnosis bias of artificial intelligence algorithms applied to chest radiographs in under -served patient populations. Nature Medicine, 27(12), 2176-2182.
- [15] Wang, S., Liu, T. Y., Wang, C., & Liu, Y. Y. (2024). Exploring the capabilities and risk delineation of generative AI in medical applications. Medicine & Philosophy, 45(12), 1-5.
- [16] Li, Q., et al. (2025). Low -dose CT imaging using a regularization -enhanced efficient diffusion probabilistic model. Biomedical Signal Processing and Control, 105, 107580.
- [17] Dalla Serra, F., et al. (2025). Multimodal generation of radiology reports using knowledge -grounded extraction of entities and relations. In Proceedings of the Association for Computational Linguistics (ACL) (pp. 615-624).
- [18] Alharbi, A. O. (2026). Automated radiology report generation using multimodal foundation models: A systematic review of clinical accuracy and safety. The Review of Diabetic Studies, 12(1).
- [19] Chen, S. (2025). A review of multimodal large model based medical image report generation. Journal of Medical Artificial Intelligence, 9(2), 248-255.
Article Sidebar
This work is licensed under a Creative Commons Attribution 4.0 International License.