Spatial transcriptomics and artificial intelligence: a scoping review of emerging applications in head and neck pathology

Single cell spatially resolved transcriptomics (ST) has revolutionized molecular profiling by providing the visualization of gene expression within its native tissue architecture, enabling insights into cellular heterogeneity, tumor microenvironment (TME) composition, and the molecular pathways driving disease progression. At the same time, advances in artificial intelligence (AI)-driven workflows have demonstrated significant applications within the medical field and are expected to transform the way complex diagnostic and prognostic challenges are approached. In addition, integrative analyses of spatial, histological and molecular data, offer new opportunities to uncover driver genes, identify new immunohistochemical biomarkers, and inform personalized treatment strategies, ultimately contributing to enhanced clinical decision-making and improved patient outcomes. This scoping review aims to examine recent research leveraging AI in ST to study head and neck (H&N) pathology and highlight future applications of these technologies for improving the diagnosis, risk stratification, and malignant transformation prediction. Scoping literature review was conducted in accordance with PRISMA guidelines using seven electronic databases, including PubMed, Embase, Cochrane Library, IEEE Xplore, EBSCOhost, Springer, and Google Scholar. Database-specific search strategies and manual reference screening were applied to identify relevant studies published between January 2014 and May 2025. Ten relevant studies were included in this review after removal of duplicates and exclusion of irrelevant articles due to incompatible formats, lack of spatial transcriptomics data, not including head and neck human tissue, or unavailable full-text access. This review identifies a substantial gap in the application of ST and AI within H&N pathology. Future research should focus on developing multimodal, AI-driven frameworks that integrate histopathology, spatial gene expression, and clinical metadata to improve early detection, risk stratification, and clinical decision-making in the management of OPMDs. Broader adoption of these approaches is essential to advance translational research and improve patient outcomes.