Comparison between Manual Lateral Cephalometric Analysis and Artificial Intelligence Driven Platforms

Introduction: It remains uncertain whether artificial intelligence based programs can detect cephalometric landmarks with accuracy. Therefore, the aim of this prospective study was to compare cephalometric measurements obtained with manual tracing and artificial intelligence programs. An additional goal was to evaluate the difference in cephalometric tracing duration.
Materials and methods: Fifty pretreatment lateral cephalometric radiographs with complete permanent dentition were included in this study. Poor quality radiographs, irregular shaped incisors roots, craniofacial deformity and/or impacted teeth were excluded. Each lateral cephalogram was traced and analysed using manual method, WebCeph and Cephio programs. After completion of landmark plotting, linear and angular measurements of Steiner’s analysis were calculated.
Results: The results indicated the presence of no statistically significant differences between the three methods. SNA, SNB and ANB angles had a relatively higher values with WebCeph compared to Cephio and manual cephalometric analysis. On the other hand, mandibular plane angle and occlusal plane angle were higher with Cephio when compared to WebCeph and manual methods. The duration taken for cephalometric measuremens showed statistically significant difference between manual cephalometric analysis and AI based programs with WebCeph presenting the least duration and manual method showing highest duration.
Conclusion: The cephalometric measurements obtained from both WebCeph and Cephio programs are highly accurate when compared to manual measurements. Cephalometric measurements done using WebCeph and Cephio are formulated in a significantly shorter time in comparison to manual method.