Artificial Intelligence for Onsite Cytology Evaluation in Endoscopic Ultrasound

What this Trial Is About

Researchers are evaluating whether the widely available artificial intelligence platform ChatGPT can accurately identify abnormalities on cytology specimens taken during endoscopic ultrasound-guided fine needle biopsy (EUS-FNB) of solid mass lesions in the gastrointestinal tract. The study compares AI interpretations with those made by experienced cytopathologists, who are considered the gold standard. The research aims to address the limited availability of onsite cytopathology services that currently affect diagnostic accuracy and efficiency. The study involves the use of EUS-guided fine needle biopsy to collect cytology specimens from solid mass lesions in the gastrointestinal tract. There are two interpretation groups: one where cytopathologists review specimens onsite, and another where the AI program analyzes the samples. The study observes and compares diagnostic accuracy between these two methods without intervening in patient care. Participants will undergo the biopsy procedure, and their cytology specimens will be evaluated either by onsite cytopathologists or by the AI program. Researchers will measure the diagnostic accuracy at rapid onsite evaluation within one day and compare it to final pathology results obtained within seven days. The study is observational and will monitor these outcomes without additional treatment, lasting until October 2027.

AI for Onsite Cytology Evaluation in Endoscopic Ultrasound