Advances in Digital Cytopathology and Artificial Intelligence Applications.

Abstract

Cytopathology is the first field of pathology in which artificial intelligence (AI) models were successfully developed and commercialized for routine clinical screening of cervical cytology, a practice that has been in place for the past 2 to 3 decades. However, the development and deployment of AI applications for nongynecologic cytology has just begun. The variety of cytology specimen types and preparations with associated unique characteristics presents technical challenges for the complete digitization of the cytology workflow. Despite of these challenges, a few institutions have adopted a complete digital cytology workflow. Technical advancement in digital cytopathology have replaced conventional rapid onsite evaluation by a variety of virtual telecytology systems. Novel digital diagnostic solutions for cytology are evolving. Among these, Hologic Genius is the only one approved by the Food and Drug Administration (FDA) for routine clinical screening of cervical cytology in the United States. The recommendations for AI validation and best-practice guidelines for digital cytopathology are currently being developed. Prospect of technical and AI advances in digital cytopathology include automation of sample preparation, ROSE using telecytology, automation of screening of gynecologic and nongynecologic cytology specimens, automated quantitation of biomarkers, quality control, and beyond. This review article uncovers recent advances in digital cytopathology and discusses potential use cases of AI applications for routine cytopathology practice in this modern era of digital cytopathology.