Related Experiment Video
Updated: Jun 3, 2026
08:21
Point-of-Care Ultrasound: A Review of Ultrasound Parameters for Predicting Difficult Airways
Published on: April 7, 2023
A two-step deep learning framework for predicting difficult video laryngoscopy from ultrasound images: a prospective
Chenyu Jin1, Bei Pei1, Ren Zhou1
1Department of Anesthesiology, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China.
BMC Anesthesiology
|June 2, 2026
Summary
AI-powered ultrasound accurately predicts difficult video laryngoscopy (DVL). The distance from skin to tongue root (DSTR) is a key predictor, with higher DSTR values significantly increasing DVL risk.
Area of Science:
- Anesthesiology
- Medical Imaging
- Artificial Intelligence
Background:
- Difficult airway management is a critical anesthetic challenge.
- Ultrasound and deep learning offer potential for predicting difficult airways.
- This study investigated AI-based ultrasound for predicting difficult video laryngoscopy (DVL).
Purpose of the Study:
- To assess the feasibility of AI-based ultrasound in predicting DVL.
- To identify specific ultrasonographic measurements associated with DVL using AI heatmaps.
Main Methods:
- Patients undergoing airway ultrasound examinations were enrolled.
- ResNet-18 and machine learning algorithms (LightGBM) were used for predictive modeling.
- AI-generated heatmaps and logistic regression identified key ultrasonographic predictors.
Main Results:
- The AI model achieved an AUROC of 0.804 for DVL prediction.
- Distance from skin to tongue root (DSTR) was significantly associated with DVL (OR: 4.83).
- A nonlinear association was observed, with increased DVL risk at DSTR ≥ 3.89 cm.
Conclusions:
- AI-based ultrasound technology demonstrates feasibility for predicting DVL.
- DSTR is a significant ultrasonographic predictor of DVL.