The peritumoral region harbors critical microenvironmental heterogeneity information and has long been a focal point in radiomics research. This article systematically reviews the progress of multicenter peritumoral radiomics in oncology, with emphasis on its clinical and biological applications. Studies indicate that peritumoral radiomics may demonstrate value in tumor diagnosis/differential diagnosis, molecular subtyping identification, prediction of clinicopathological indicators, prognostic risk assessment, and treatment response evaluation. Multicenter validation shows that incorporating peritumoral features can enhance model performance for certain tasks, with peritumoral characteristics potentially outperforming intratumoral features in specific scenarios. At the biological mechanism level, peritumoral radiomics exhibits associations with genomics and immune microenvironment events, offering novel perspectives for non-invasive tumor biology analysis. However, the field still faces challenges including limited model generalizability, standardization and reproducibility issues, demands for interpretability, ambiguous peritumoral region definitions, and multi-omics integration barriers - these limitations necessitate cautious interpretation of current findings. Future breakthroughs may rely on prospective multicenter studies, standardized protocols, explainable AI techniques, biologically-driven optimization of peritumoral definitions, and innovative multi-omics fusion algorithms. To promote clinical translation, it is necessary to enhance model robustness and interpretability while deepening research on the association with tumor biological mechanisms.
Li et al. (Mon,) studied this question.