Related Experiment Video
Updated: Jul 18, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.8K
Geometric morphometrics and machine learning from three-dimensional facial scans for difficult mask ventilation
Bei Pei1, Chenyu Jin1, Shuang Cao1
1Department of Anaesthesiology, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China.
Frontiers in Medicine
|August 28, 2023
Summary
Three-dimensional facial scans can predict difficult mask ventilation (DMV) in anesthesia. Machine learning models, particularly logistic regression, showed high accuracy, identifying mandibular morphology as a key predictor.
Area of Science:
- Anesthesiology
- Medical Imaging
- Machine Learning
Background:
- Difficult mask ventilation (DMV) poses a significant risk during anesthesia, and its prediction remains challenging.
- Current methods for predicting DMV have limitations, necessitating novel approaches.
- This study investigates the potential of 3D facial scans for predicting DMV.
Purpose of the Study:
- To evaluate the efficacy of 3D facial scans in predicting difficult mask ventilation (DMV) in patients undergoing general anesthesia.
- To compare the predictive performance of machine learning models utilizing 3D facial data against existing clinical scores.
- To identify specific facial morphological features associated with DMV.
Main Methods:
- 669 adult patients scheduled for elective surgery under general anesthesia were included.
- 3D facial scans were analyzed using 3D geometric morphometrics to capture detailed facial features.
- Ten machine learning algorithms were employed to predict DMV, with performance assessed using AUC, sensitivity, and specificity.
Main Results:
- The incidence of DMV was 5.23%.
- The logistic regression model achieved the highest predictive performance with an AUC of 0.825.
- Significant morphological differences in the mandibular region were observed between patients with and without DMV.
Conclusions:
- 3D geometric morphometrics combined with machine learning offers a promising non-invasive method for predicting DMV.
- The logistic regression model demonstrated superior predictive accuracy compared to the DIFFMASK score.
- Identifying distinct mandibular morphology in patients prone to DMV can enhance anesthesia safety.
