Deep‐learning model associating lateral cervical radiographic features withCormack–Lehanegrade 3 or 4 glottic view

Abstract

Summary Unanticipated difficult laryngoscopy is associated with serious airway‐related complications. We aimed to develop and test a convolutional neural network‐based deep‐learning model that uses lateral cervical spine radiographs to predict Cormack–Lehane grade 3 or 4 direct laryngoscopy views of the glottis. We analysed the radiographs of 5939 thyroid surgery patients at our hospital, 253 (4%) of whom had grade 3 or 4 glottic views. We used 10 randomly sampled datasets to train a model. We compared the new model with six similar models (VGG, ResNet, Xception, ResNext, DenseNet and SENet). The Brier score (95%CI) of the new model, 0.023 (0.021–0.025), was lower (‘better’) than the other models: VGG, 0.034 (0.034–0.035); ResNet, 0.033 (0.033–0.035); Xception, 0.032 (0.031–0.033); ResNext, 0.033 (0.032–0.033); DenseNet, 0.030 (0.029–0.032); SENet, 0.031 (0.029–0.032), all p < 0.001. We calculated mean (95%CI) of the new model for: R2, 0.428 (0.388–0.468); mean squared error, 0.023 (0.021–0.025); mean absolute error, 0.048 (0.046–0.049); balanced accuracy, 0.713 (0.684–0.742); and area under the receiver operating characteristic curve, 0.965 (0.962–0.969). Radiographic features around the hyoid bone, pharynx and cervical spine were associated with grade 3 and 4 glottic views.

A review of the current status and progress in difficult airway assessment research

“…In addition to the direct collection of patient's facial images for difficult airway prediction by deep learning, there are some studies that analyze the radiographic pictures of patient's head, face, and neck for difficult airway prediction by deep learning. For example H-Y CHO et al developed a model for predicting a difficult airway based on convolutional neural network algorithm by using patient's cervical spine lateral X-ray images [ 62 ]. The emergence of these models and algorithms may provide a new way of thinking about the clinical assessment of difficult airways.…”

Opportunities of AI-powered applications in anesthesiology to enhance patient safety

“…27 In another study in which lateral cervical spinal radiographs were available, a convolutional neural network model developed using images from the same institution predicted the Cochrane-Lehane grade 3 and 4 glottic views with an AUC of 0.96. 28 These early results are promising, and accurately predicting the patients at risk for difficult airways would allow better preparedness and decrease complications. To implement this technology, future studies need to develop highly performant models relevant to real-world, diverse clinical practice.…”

“…Deep learning models utilizing convolutional neural networks have also been used successfully in another single-center study; however, the above concerns about study design and validation remain 27 . In another study in which lateral cervical spinal radiographs were available, a convolutional neural network model developed using images from the same institution predicted the Cochrane-Lehane grade 3 and 4 glottic views with an AUC of 0.96 28 . These early results are promising, and accurately predicting the patients at risk for difficult airways would allow better preparedness and decrease complications.…”

Automated location of orofacial landmarks to characterize airway morphology in anaesthesia via deep convolutional neural networks

“…In addition, Hayasaka et al [20] proposed a methodology based on photos of 16 different poses per patient (combinations of frontal/lateral views, supine/sitting position, open/closed mouth, head bent backwards or not), training respective CNNs to classify difficult airways. Furthermore, Cho et al [21] used CNNs with lateral cervical spine X-ray images.…”