Background
Tracheal intubation is the gold standard for securing the airway, and it is not uncommon to encounter intubation difficulties in intensive care units and emergency rooms. Currently, there is a need for an objective measure to assess intubation difficulties in emergency situations by physicians, residents, and paramedics who are unfamiliar with tracheal intubation. Artificial intelligence (AI) is currently used in medical imaging owing to advanced performance. We aimed to create an AI model to classify intubation difficulties from the patient’s facial image using a convolutional neural network (CNN), which links the facial image with the actual difficulty of intubation.
Methods
Patients scheduled for surgery at Yamagata University Hospital between April and August 2020 were enrolled. Patients who underwent surgery with altered facial appearance, surgery with altered range of motion in the neck, or intubation performed by a physician with less than 3 years of anesthesia experience were excluded. Sixteen different facial images were obtained from the patients since the day after surgery. All images were judged as “Easy”/“Difficult” by an anesthesiologist, and an AI classification model was created using deep learning by linking the patient’s facial image and the intubation difficulty. Receiver operating characteristic curves of actual intubation difficulty and AI model were developed, and sensitivity, specificity, and area under the curve (AUC) were calculated; median AUC was used as the result. Class activation heat maps were used to visualize how the AI model classifies intubation difficulties.
Results
The best AI model for classifying intubation difficulties from 16 different images was generated in the supine-side-closed mouth-base position. The accuracy was 80.5%; sensitivity, 81.8%; specificity, 83.3%; AUC, 0.864; and 95% confidence interval, [0.731-0.969], indicating that the class activation heat map was concentrated around the neck regardless of the background; the AI model recognized facial contours and identified intubation difficulties.
Conclusion
This is the first study to apply deep learning (CNN) to classify intubation difficulties using an AI model. We could create an AI model with an AUC of 0.864. Our AI model may be useful for tracheal intubation performed by inexperienced medical staff in emergency situations or under general anesthesia.
Adequate modeling of light propagation in the head is important to predict the
sensitivity of NIRS signal and the spatial sensitivity profile of
source-detector pairs. The 3D realistic head models of which the geometry is
based upon the anatomical images acquired by magnetic resonance imaging and
x-ray computed tomography are constructed to investigate the influence of the
frontal sinus on the NIRS signal and spatial sensitivity. Light propagation in
the head is strongly affected by the presence of the frontal sinus. The light
tends to propagate around the frontal sinus. The influence of the frontal sinus
on the sensitivity of the NIRS signal to the brain activation is not consistent
and depends on the depth of the frontal sinus, the optical properties of the
superficial tissues and the relative position between the source-detector pair
and the frontal sinus. The frontal sinus located in the shallow region of the
skull tends to reduce the sensitivity of the NIRS signal while the deep frontal
sinus can increase the sensitivity of the NIRS signal.
Subject-specific head models of which their geometry is based on structural magnetic resonance images are essential to accurately estimate the spatial sensitivity profiles for image reconstruction in diffuse optical tomography. T1-weighted magnetic resonance images, which are commonly used for structural imaging, are not sufficient for the threshold-based segmentation of the superficial tissues. Two types of pulse sequences, which provide a high contrast among the superficial tissues, are introduced to complement the segmentation to construct the subject-specific head models. The magnetic resonance images acquired by the proposed pulse sequences are robust to the threshold level and adequate for the threshold-based segmentation of the superficial tissues compared to the T1- and T2-weighted images. The total scan time of the proposed pulse sequences is less than one-fourth of that for the T2-weighted pulse sequence.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.