Pivotal clinical study to evaluate the efficacy and safety of as-sistive artificial intelligence-based software for cervical cancer diagnosis

초록

Introduction The accuracy of colposcopy depends on the skill and proficiency of the colposcopist. This study evaluated the feasibility of an artificial intelligence (AI) system as an assistive tool for diagnosing high-grade cervical intraepithelial neoplasia lesions compared to the human interpretation of cervical images Methods This two-centered, crossover, double-blind, randomized controlled trial included 886 randomly selected images. Four colposcopists (two proficient and two inexperienced) independently evaluated the cervical images once with and without the aid of the Cerviray AI® system (AIDOT, Seoul, Korea). Results The AI aid demonstrated improved areas under the curve on the localization receiver-operating characteristic curve compared with the colposcopy impressions of colposcopists (difference 0.12, 95% confidence interval [CI], 0.10 – 0.14, p-value < 0.001). Sensitivity and specificity also improved on using AI system (89.18% vs. 71.33%; p < 0.001, 96.68% vs. 92.16%; p < 0.001, respectively). Additionally, the classification accuracy rate improved with the aid of AI (86.40% vs. 75.45%; p < 0.001). Conclusion/Implications This study highlights the feasibility of using an AI system as an effective assistive tool for both proficient and inexperienced colposcopists in cervical cancer screening. AI interpretation can be used as an assisting tool in combination with human colposcopic evaluation of the exocervix.