[Artificial intelligence-driven assessment of tertiary lymphoid structures in breast cancer: From microscopic analysis to clinical translation].

Tertiary lymphoid structures (TLSs) are key regulatory components of the breast cancer immune microenvironment. Their microscopic composition (specific immune cell subsets, spatial organization) and functional status (maturity) directly determine their performance in coordinating the anti-tumor immune responses, serving as the microscopic foundation for understanding the efficacy of immunotherapy. However, the high heterogeneity and dynamic evolution of the tumor microenvironment pose significant limitations for immunotherapeutic efficacy prediction models based on traditional histopathological staging and molecular subtyping. In this context, developing non-invasive assessment technologies capable of quantitatively analyzing the spatial distribution and functional activity of TLSs holds crucial clinical value for achieving precise immunotherapy response prediction and prognostic stratification. Artificial intelligence (AI) technology, particularly the deep learning algorithms integrating multi-omics data, offers innovative tools for systematically decoding TLSs by leveraging its unique advantages in complex feature extraction and high-dimensional data analysis. This review focuses on how AI technology deciphers the microcosmic nature of TLSs. It systematically summarizes recent advances in AI-driven analysis of multi-modal data, including genomics, pathological images, and medical imaging, to decode TLSs. It also delves into the challenges lying in these technologies (such as data standardization and model interpretability) and envisions future pathways for advancing TLS research from microscopic insights to personalized precision immunotherapy.