Evaluation of Pharyngeal Airway Volume in Patients with Impacted Maxillary Canines Using Cone-beam Computed Tomography and Comparative Analysis of Artificial Intelligence-Based Segmentation Methods.
Authors
Affiliations (2)
Affiliations (2)
- Private Clinic, Faculty of Dentistry, İnönü University, Malatya, Turkey.
- Department of Orthodontics, Faculty of Dentistry, İnönü University, Malatya, Turkey.
Abstract
Impacted maxillary canines may be associated with alterations in adjacent craniofacial structures; however, their relationship with pharyngeal airway dimensions has not been directly investigated. The aim of this study was to compare pharyngeal airway volume between individuals with impacted maxillary canines and a control group, and to evaluate the measurement outcomes obtained using semi-automatic and artificial intelligence (AI)-based segmentation methods. Cone-beam computed tomography data from 120 retrospectively selected individuals were analyzed. Pharyngeal airway volumes were measured using a semi-automatic method (NemoFAB), and comparisons were made between the impacted canine group ( n = 70) and the control group ( n = 50). In the second phase, NemoFAB measurements from 95 patients were compared with those obtained from four different AI-based software programs. Non-parametric tests, analysis of covariance (ANCOVA), and correlation analyses were performed for statistical evaluation. Airway volume, minimum cross-sectional area, and axial slice area values were significantly higher in the control group compared with the impacted canine group ( P ≤ 0.005). The group effect remained statistically significant after controlling for age and sex ( P < 0.001). Although statistically significant differences were observed between the AI-based programs and the semi-automatic method, high correlation coefficients ( r = 0.903-0.947) and strong intraclass correlation coefficient values indicated a high level of agreement between the methods. The presence of impacted maxillary canines is associated with a reduction in pharyngeal airway volume. AI-based segmentation methods demonstrate a high level of agreement with the semi-automatic method and appear to be suitable for clinical and research applications.