Advances in digital pathology and artificial intelligence (AI) are significantly transforming the approach to analyzing the tumor microenvironment (TME) in gastrointestinal cancers (GICs). The TME consists of tumor cells, stromal components, and immune cells. It plays a key role in disease progression, treatment response, and patient prognosis. This review discusses the most important TME biomarkers, such as tumor budding (TB), tumor-infiltrating lymphocytes (TILs), and tertiary lymphoid structures (TLSs), with emphasis on their prognostic and predictive significance. Traditional histopathological assessment of these parameters is limited by subjectivity, intraobserver variability, and time-consuming nature. In this context, AI-based tools enable automated, quantitative, and more reproducible analysis of entire histological sections. Deep learning models allow the accurate detection and classification of structures and also analysis of their spatial organization. They provide new biological insights unavailable in routine diagnostics. The integration of imaging data with molecular and clinical information leads to the development of personalized medicine. Despite numerous advantages, the implementation of AI in clinical practice continues to face challenges related to standardization, data availability, and model interpretability.
Łapińska et al. (Thu,) studied this question.