Artificial Intelligence System to Detect Lesions and Recognize Anatomy Using Nasopharyngolaryngoscopy Images in Adults A Prospective Multicenter Study

What this Trial Is About

Researchers are evaluating an artificial intelligence-assisted system designed to improve the detection of anatomical sites and lesions during nasopharyngolaryngoscopy, a procedure used to examine the nasopharynx and larynx. The study aims to address challenges of traditional procedures such as incomplete visualization and unclear imaging by training and validating a deep learning model using collected images and patient data from multiple centers. The study involves collecting electronic nasopharyngolaryngoscopy images and baseline patient information, including gender and age, to train and validate the AI model. Initially, the model is trained using a retrospective dataset and internally validated. Subsequently, the clinical performance of the model is assessed prospectively by enrolling patients, collecting new nasopharyngolaryngoscopy videos, and testing the model against this independent dataset. Participants will undergo standard nasopharyngolaryngoscopy exams, with their images and videos collected for analysis. Researchers will compare the models lesion detection and anatomic site recognition performance to that of physicians. Outcome measures are evaluated within three months after completing data collection. The study runs from December 2025 to March 2027 and includes assessments of diagnostic accuracy and comparative performance between the AI system and medical professionals.

AI System for Anatomic Recognition & Lesion Detection in Nasopharyngolaryngoscopy: A Prospective Study