肿瘤 z-bio 2025-03-06 387

　　HistoCell algorithm framework (a) and its prediction accuracy for cell type information related to tumor pathological imaging (b)

　　With the support of the National Natural Science Foundation of China project (Approval No. T2341008) and other grants, Professor Li Shao's research group from the Beijing Institute of Traditional Chinese Medicine at Tsinghua University has made research progress in intelligent analysis of microscopic information in pathological images of both Chinese and Western medicine, promoting precise prevention and treatment of tumors. The research results, titled "Systematic inference of super-resolution cell spatial profiles from history images", were published online on February 21, 2025 in the journal Nature Communications. The paper link is: https://www.nature.com/articles/s41467-025-57072-6 .

　　Decoding the correlation between pathological imaging features and clinical macroscopic phenotype, microscopic cellular information through artificial intelligence algorithms, and revealing the diagnosis and treatment rules of complex diseases such as tumors in traditional Chinese and Western medicine, is currently a research hotspot in the field of medical imaging. This study proposes a new algorithm for inferring the relationship between pathological images and cell networks based on a weakly supervised learning framework - HistoCell. This algorithm comprehensively characterizes the pathological morphological features and spatial topological features, combined with the hierarchical encoding rules embedded at the cellular level, to achieve spatial correlation network recognition of pathological microscopic information at the single-cell scale, expanding the scope of analysis of pathological imaging related microscopic information. The research team applied the HistoCell algorithm to early warning of gastric cancer, prognostic risk analysis of breast cancer and other cancer chemotherapy drug response prediction and other clinical diagnosis and treatment scenarios, and verified its application potential in data mining of traditional and western medicine imageomics and promoting precise prevention and treatment of complex diseases.