A universal visual foundation model for computational cytopathology

Researchers developed CROWN, a universal visual foundation model for computational cytopathology, and benchmarked it across more than 200 real-world cytology tasks spanning multiple cohorts, including lymph node metastasis detection and cervical screening datasets. The model demonstrates broad applicability as a single pretrained architecture capable of handling diverse cytological classification and screening challenges. Published in Nature Cancer, CROWN represents a significant step toward generalizable AI-driven cytopathology, though clinical deployment and prospective validation remain to be demonstrated.