Development of a semi-automatic collaborative system for cephalometric measurement in orthodontic radiographic images

Cephalometric analysis is a fundamental tool in orthodontics for evaluating craniofacial morphology using radiographic images. However, traditional manual methods are time-consuming and show inter-observer variability, while fully automated artificial intelligence systems still face limitations in clinical reliability and generalization. This study presents a collaborative semi-automatic system designed to improve efficiency and consistency in two-dimensional cephalometric analysis. The system integrates digital templates with expert-assisted landmark identification, allowing trained users to mark anatomical reference points while the software automatically performs geometric calculations. A comparative evaluation was conducted between the manual and semi-automatic methods, focusing on processing time, landmark localization accuracy, and operational workload. The results showed a significant reduction in average processing time per radiograph, decreasing from 12.5 minutes using the manual method to 4.3 minutes with the proposed system. In addition, a decrease in variability in landmark identification was observed, particularly in mandibular structures where manual measurements tend to be less consistent. The semi-automatic approach also reduced workload by automating calculation and data recording processes while maintaining expert control during the marking phase. It is concluded that the combination of clinical expertise and computational support improves both efficiency and consistency in orthodontic analysis.