Agreement and reliability of AI-based cephalometric analysis (WebCeph) compared with digital manual tracing for skeletal and dentoalveolar measurements (A retrospective comparative study)

Background: Cephalometric analysis is essential in orthodontic diagnosis and treatment planning. With the increasing use of artificial intelligence (AI) in clinical practice, automated platforms such as WebCeph have been introduced; however, their accuracy and reliability across different types of measurements remain a concerns. Objectives: evaluate the agreement and reliability between skeletal and dentoalveolar cephalometric measurements obtained using an AI-based platform (WebCeph) and digital manual tracing performed with AutoCAD.Materials and Methods: A total of 100 lateral cephalograms were analyzed using WebCeph (AI-driven software) and digital manual tracing in AutoCAD. Fourteen Steiner and Tweed variables were selected and categorized into skeletal and dentoalveolar measurements. Statistical comparisons were performed using paired t-tests or Wilcoxon signed-rank tests, depending on data distribution. Agreement between the two methods was assessed using intraclass correlation coefficients (ICC) and Bland–Altman analysis. Conclusion: Overall agreement between WebCeph and AutoCAD was good for most variables (ICC ≥ 0.75). Skeletal and dentoalveolar measurements showed comparable reliability, although skeletal measurements demonstrated slightly smaller mean differences and narrower limits of agreement. Greater variability was observed in dentoalveolar variables, particularly those related to mandibular incisor inclination. Conclusion:WebCeph provides clinically reliable cephalometric measurements and can be considered a supportive diagnostic tool. However, selective manual verification is essential.