Artificial Intelligence in Cytological Diagnosis: A Narrative Review of Current Applications, Challenges, and Future Perspectives

Background:Artificial intelligence (AI) is rapidly transforming cytological diagnostics through automated image analysis, enhanced sensitivity and specificity, and workflow efficiency in screening programs, particularly for cervical cancer. Objective:To provide a narrative overview of current AI applications in cytology, emphasizing diagnostic performance, technical limitations, and laboratory implications. Methods:A literature review was performed using PubMed, Scopus, and Google Scholar, focusing on publications from 2020–2024 with keywords such as artificial intelligence, cytology, digital pathology, and deep learning. Selected studies were synthesized to highlight real‑world implementations. Results:AI tools, especially deep learning models, have shown improved detection of abnormal cellular morphology, reduced interobserver variability, and more consistent diagnostic outcomes. Most evidence centers on cervical cytology. Remaining challenges include dataset quality, algorithmic bias, regulatory gaps, and the need for specialized training for cytopathology staff. Conclusion:AI offers promising advances in cytological diagnosis but requires validation, standardization, and workforce adaptation. Future interdisciplinary research and collaboration are essential to embed these technologies safely and ethically into routine practice.