Artificial Intelligence in Diagnostic Pathology: A Narrative review on Current Applications, Emerging Challenges, and Future Perspectives

Abstract

Advancements in digital pathology and whole-slide imaging have enabled Artificial Intelligence (AI) to emerge as a transformative tool in diagnostic pathology. AI has enabled the automated analysis of histopathological images, improving diagnostic accuracy and reproducibility, as well as workflow efficiency. The aim of this narrative review is to provide a comprehensive overview of current applications of AI in diagnostic pathology, including tumour detection, grading, biomarker prediction, and subspecialtyspecific applications, clinical utility, technical advances, along with challenges and future perspectives. The prediction of molecular alterations and therapeutic response directly from histopathological images is also made possible by AI. Cross-domain prediction and precision oncology have been further advanced by evolving technologies such as multimodal AI and foundation models. However, existing challenges such as data heterogeneity, annotation bias, lack of standardisation, limited interpretability, and regulatory constraints restrict its widespread clinical implementation. Although AI has shown significant promise in improving diagnostic accuracy, its optimal utilisation lies in a collaborative framework in which AI serves as a decision-support tool rather than a replacement for pathologists. Emerging approaches such as Explainable AI (XAI), foundation models, and multimodal AI integrating histopathology with genomic and clinical data are expected to enhance clinical translation and precision oncology. Future research focusing on large, multi-institutional datasets, prospective validation, and integration of multimodal data is needed to enable reliable clinical acceptance.