COMPUTER VISION-BASED AIRWAY ASSESSMENT USING FACIAL AND NECK IMAGING FOR PREDICTING DIFFICULT INTUBATION IN SURGICAL ANESTHESIA

Abstract

This study aims to investigate the feasibility of using computer vision facial/neck imaging for prediction of difficult intubation during surgical anesthesia. It is the prediction of a difficult airway that continues to be a significant concern during perioperative patient care as the traditional bedside airway assessment tools including Mallampati classification, thyromental distance and visual inspection are subjective, inconsistent and fails to identify patients with complex airway anatomy. The study suggests an artificial intelligence-based framework that leverages deep learning, multimodal anatomical feature extraction in cervical and facial imaging to enhance the pre-operative airway risk assessment. The technique involves the automatic analysis of facial morphology, mandibular morphology, neck morphology, cervical profile and ultrasound parameters of the airway to detect patients who may be at risk of difficult laryngoscopy or intubation. The model combines computer vision with clinical and anthropometric parameters to offer a more objective and reproducible approach to the evaluation of airway. Overall, the study underlines the potential of convolutional neural networks, vision transformers, feature-fusion models, and explainable AI techniques like SHAP in enhancing diagnostic accuracy and clinical interpretability. The reviewed methodology and discussion indicate that the use of multimodal AI-based assessment could surpass the performance of unilinear clinical predictors in capturing the complex dynamic relationships between soft tissue dimensions, cervical anatomy, and laryngoscopy difficulty. This type of approach can help anesthesiologists make decisions for appropriate airway management strategies, such as preparing for video laryngoscopy or advanced airway support as early as possible. In summary, computer vision provides a promising route to safer, quicker and more standardized decisions throughout surgical anesthesia.