Cellular-level information is routinely used as part of clinical diagnostics1. Laboratory workers use a microscope to inspect cells on glass slides to detect early signs of cancer in organs such as the lung and bladder. Although this method, called cytology, is a long-established diagnostic procedure, it has major weaknesses. Human interpretation is subjective and labour intensive, and workforce shortages are increasingly a challenge2. Writing in Nature, Nitta et al.3 present an artificial-intelligence approach that offers advantages over current methods in clinical cytology.
doi: https://doi.org/10.1038/d41586-026-00288-3
References Turner, S. A., Abou Shaar, R. & Yang, Z. Diagn. Cytopathol. 51, 83–94 (2023). Huang, Z. et al. Nature Biomed. Eng. 9, 455–470 (2025). Nitta, N. et al. Nature https://doi.org/10.1038/s41586-025-10094-y (2026). Satturwar, S. et al. J. Am. Soc. Cytopathol. 14, 65–77 (2025). Moriarty, A. T. et al. Arch. Pathol. Lab. Med. 138, 1182–1185 (2014). McAlpine, E. D., Pantanowitz, L. & Michelow, P. M. Acta Cytol. 65, 301–309 (2021). Kim, K. et al. npj Imaging 2, 39 (2024). Nitta, N. et al. Cell 175, 266–276 (2018). Ding, T. et al. Nature Rev. Bioeng. 3, 890–907 (2025). Download references
Competing Interests The authors declare no competing interests.
Related Articles
Subjects