ARTIFICIAL INTELLIGENCE AND WHOLE-SLIDE IMAGING IN CYTOPATHOLOGY: TECHNICAL CONSIDERATIONS, EVIDENCE, AND WORKFLOW INTEGRATION

Abstract

Objective: To review the technical challenges, published evidence, validation considerations, and practical workflow integration of Whole-Slide Imaging (WSI) and Artificial Intelligence (AI) in gynecologic and non-gynecologic cytopathology.

Methods: A narrative review was undertaken using peer-reviewed literature on digital cytopathology, WSI, AI-assisted cytology, validation studies, and standardized reporting systems. Emphasis was placed on reported study populations, diagnostic performance, validation approaches, workflow effects, and limitations of current models.

Results: While digital histopathology has achieved widespread clinical adoption, digital cytopathology faces distinct physical constraints, including three-dimensional cellular overlapping, variable depth-of-field, liquid-based preparation variations, and massive digital file sizes. Recent developments in multi-layer z-stacking, fast volume scanning, and deep learning models have significantly advanced automated cell detection and classification across standardized reporting schemes (including Bethesda, Paris, Milan, Sydney, and TIS). However, issues regarding pre-analytic standardization, scanner focus tracking, laboratory information system (LIS) interoperability, and validation frameworks remain critical barriers to full routine deployment.

Conclusion: AI-assisted digital cytopathology demonstrates technological feasibility and promising research-level diagnostic performance, with selected applications showing emerging clinical utility. However, routine clinical implementation remains dependent on specimen and scanner standardization, laboratory-specific validation, interoperability, infrastructure, regulatory requirements, and continued cytopathologist oversight.