Artificial intelligence (AI) and machine learning (ML) are transforming diagnostic medicine, particularly in pathology, where image-based interpretation is central to clinical decision-making. This systematic review aimed to examine recent advances, performance outcomes, and practical challenges associated with incorporating AI and ML into diagnostic pathology. A comprehensive literature search was conducted across major scientific databases for studies published between 2010 and 2025. Titles and abstracts were screened independently, full texts were assessed against predefined eligibility criteria, and data were extracted using standardised procedures. Methodological quality was evaluated using established critical appraisal tools appropriate to study design, with structured risk-of-bias assessment reported for diagnostic accuracy studies. A total of 13 studies fulfilled the inclusion criteria, covering multiple pathological domains including breast pathology, cytopathology, neuropathology, head and neck oncology, and multi-organ computational pathology. Across the included studies, deep learning approaches demonstrated high diagnostic performance for tumour detection, classification, and staging tasks. While several investigations incorporated external validation, most were retrospective in design and relied on secondary datasets. Risk-of-bias assessment indicated predominantly moderate overall risk, primarily related to study design and applicability concerns. The evidence suggests that AI and ML systems demonstrate strong technical performance in controlled validation settings and may function as assistive tools within digital pathology workflows.
Mishra et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: