Evaluation of the pharyngeal airway with artificial intelligence algorithms improved by deep learning method on cone-beam computed tomography images

Aim: The aim of this study is to investigate the success of pharyngeal airway detection using a special artificial intelligence algorithm on lateral cephalometric images obtained from cone beam computed tomography images. Material and Method: The data set of our study was performed on the lateral cephalometric radiographs obtained from cone beam computed tomography images of 1040 patients before orthodontic treatment using a special artificial intelligence algorithm and the segmentation method was applied with yhe free drawing tchnique and the pharyngeal airway was determined. Airway labeling on images was done using CranipCatch software (Craniocatch, Eskişehir, Turkey). Results: The artificial intelligence model was trained with the Yolov5 Segmentation Yolov5x model as 500 epochs. Sensitivity, precision and F1 scores in the artifical intelligence model trained in the study were 1, 0.9903846154 and 0.9951690821 respectively. The learning rate of the model was found to be 0.01. Conclusion: There were no missing tags in our study, and the model was generally successful. Our study is very important for the development of deep learning-based CBCT reporting systems to be made in the future. It is thought that these systems will play a role as a decision support mechanism by saving time for physicians in their clinical routine. At the same time, it is estimated that during the determination of the pharyngeal airway, it will be very helpful in minimizing the differences between observers and the inconsistencies that may occur in the evaluations of the observers at different times. Keywords: Artificial Intelligence, CBCT, Deep learning, detection, the pharyngeal airway