Difficult airway is a life-threatening event during anesthesia. Prediction model is helpful to detect high-risk patients and decrease the risk of un-anticipated difficult airway. Present models are usually based on Mallampati grade and the width of mouth open. However, the prediction accuracy is only about 0.7-0.8 in different populations. Present study is designed to investigate if AI-based prediction model using medical imaging parameters (such as CT and MRI) can increase the accuracy of prediction model.

Inclusion Criteria:

1. age ≥18 years old;
2. surgical patients undergoing general anesthesia with endotracheal intubation;
3. with head and neck CT examination results
4. Consent to participate in the study.

Exclusion Criteria:

1. The presence of laryngeal edema;
2. The presence of airway stenosis, including internal airway stenosis (such as foreign body or tumor) or stenosis caused by external tracheal mass compression;
3. tracheo-esophageal fistula;
4. severe gastroesophageal reflux;
5. previous upper airway surgery, such as laryngeal cancer radical surgery, snoring surgery, etc.

6）participating in other research projects

Patients with head and neck CT data undergoing surgery under general anesthesia with endotracheal intubation