Accuracy and Clinical Applicability of Automated 3D CBCT-based Cephalometric Analysis in Children and Adolescents

This study aimed to evaluate the accuracy and clinical applicability of an automated landmark identification (ALI) system for three-dimensional cone-beam computed tomography (CBCT)-based cephalometric analysis in pediatric and adolescent patients. The influence of dentition stage on system performance was also investigated. Sixty individuals who underwent CBCT imaging for orthodontic diagnosis were equally divided into mixed dentition and permanent dentition groups. Thirty landmarks were automatically identified using the ALI function of Invivo7 software (Anatomage, San Jose, CA, USA). For comparison, manual identification was performed by a primary examiner, and inter-examiner reliability was assessed in a random subset of 10 samples by a second independent examiner (ICC > 0.86). Landmark and measurement accuracies were evaluated as the mean error distance (MED) and mean absolute error (MAE), respectively. Successful detection and measurement rates were calculated using a predefined threshold of 2° or 2 mm. The ALI system demonstrated high repeatability and comparable accuracy to manual identification in both groups. Although some skeletal and dental landmarks showed localization errors > 2 mm, overall MED or MAE did not significantly differ between dentition stages. However, certain measurements, including posterior facial height and U1–SN, showed high variability. These findings suggest that the ALI system can be an efficient and consistent adjunctive tool for cephalometric analysis in growing patients. However, selective manual verification is required for specific landmarks and measurements with high variability.