Applicability of Artificial Intelligence Analysis in Oral Cytopathology: A Pilot Study.

Abstract

Introduction: Oral cancer, especially oral squamous cell carcinoma (OSCC), is a global health challenge due to factors such as late detection and high mortality rates. Early detection is essential through monitoring by healthcare professionals. Cytopathology is a cellular analysis model for evaluating cellular damage preceding the clinical appearance of OSCC, but it requires training and has diagnostic limitations, due to its subjective aspect. Artificial intelligence (AI) shows potential to enhance the interpretation of cytological images, reducing working time and subjectivity.

Objective: The aim of the study was to compare the effectiveness of human analyses versus AI system assessment of oral cell smears stained by the Papanicolaou technique.

Methodology: The study comprised 57 patients in Porto Alegre - RS divided into four groups: control group (CG), exposed group (EG), oral potentially malignant disorders group (OPMDG), and OSCC group (OSCCG). Cytopathological smears were collected from the border of the tongue of CG and EG and from the lesional area in OSCCG and OPMDG. The Papanicolaou technique was performed according to standard protocol, with morphological analysis. Images were analyzed by two human examiners as well as by an AI system (Papanicolaou Slide Image Examiner [PSIE]).

Results: Concordance between human and PSIE was good. The proportion of cytological findings between human and PSIE was similar, and the analysis time of PSIE was 16.6 times shorter than that of human researchers.

Conclusion: The use of AI for OSCC screening is promising and demonstrated to be a suitable tool for routine use mainly with the advance of IA-human concordance analysis and serving as a tool to accelerate the analytical process.