Journal of clinical anesthesia · 2025
Development of an artificial intelligence-assisted system for tracheal intubation using a video laryngoscope in infants and neonates.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Nakamura Hiroki, Fukuda Kouki, Asano Takahiko, Masue Tatsuhiko
Аннотация
BACKGROUND: The larynx of infants and neonates is occasionally challenging to identify using video laryngoscope. This study aimed to develop an artificial intelligence (AI)-assisted model that can identify the larynx, including the opening vocal cord and the arytenoid portion, on images obtained using a video laryngoscope.
METHODS: First, 1197 images were extracted by the author from the 653 videos for train and validation data. An AI-assisted model for identifying the larynx was developed using YOLOv8n. Then, 399 images were selected from the additional 63 videos for the test data at every 150 frames.
RESULTS: The sensitivity, specificity, and area under the curve of the AI-assisted model for identifying the larynx were 0.74, 0.99, and 0.91, respectively. The AI model correctly identified the opening vocal cord and arytenoid portion in cases without obstacles. Esophageal misidentification of the larynx and undetected cases of the larynx caused by obstacles were observed.
CONCLUSIONS: The AI-assisted model effectively identified the larynx, including the opening vocal cord and the arytenoid portion, during video laryngoscopy and potentially can enhance the safety of tracheal intubation of infants and neonates. However, esophageal misidentification remains critical. Further studies are needed to refine the model and ensure its reliability in clinical practice.
METHODS: First, 1197 images were extracted by the author from the 653 videos for train and validation data. An AI-assisted model for identifying the larynx was developed using YOLOv8n. Then, 399 images were selected from the additional 63 videos for the test data at every 150 frames.
RESULTS: The sensitivity, specificity, and area under the curve of the AI-assisted model for identifying the larynx were 0.74, 0.99, and 0.91, respectively. The AI model correctly identified the opening vocal cord and arytenoid portion in cases without obstacles. Esophageal misidentification of the larynx and undetected cases of the larynx caused by obstacles were observed.
CONCLUSIONS: The AI-assisted model effectively identified the larynx, including the opening vocal cord and the arytenoid portion, during video laryngoscopy and potentially can enhance the safety of tracheal intubation of infants and neonates. However, esophageal misidentification remains critical. Further studies are needed to refine the model and ensure its reliability in clinical practice.