Web-based Fully Automated Cephalometric Analysis: Comparisons between App-aided, Computerized, and Manual Tracings

Area of Science:

• Orthodontics
• Radiology
• Artificial Intelligence in Medicine

Background:

• Cephalometric analysis is crucial in orthodontics for diagnosis and treatment planning.
• Traditional manual tracing is time-consuming and prone to inter-operator variability.
• Automated and app-aided cephalometric tracing methods offer potential for increased efficiency and accuracy.

Purpose of the Study:

• To compare the accuracy of fully automated (CephX), computerized (Dolphin), and app-aided (CephNinja) cephalometric tracings against manual tracings.
• To evaluate the time efficiency of each cephalometric analysis method.

Main Methods:

• Lateral cephalometric radiographs of 40 patients were analyzed using Dolphin, CephNinja, and CephX.
• Manual tracings were performed by three operators for comparison.
• Intra- and inter-examiner reliability (ICC) and inter-group differences (ANOVA, Tukey test) were assessed.
• Tracing time was recorded for each method.

Main Results:

• CephX exhibited greater variability and statistically significant differences in several parameters compared to manual tracings.
• CephNinja and Dolphin analyses were comparable to manual tracings.
• Manual correction of CephX landmarks improved its accuracy, making it comparable to CephNinja and Dolphin.
• CephX offered the shortest analysis time.

Conclusions:

• Fully automated CephX requires further refinement for improved reliability.
• CephX with manual landmark correction presents a promising, time-efficient option for clinical cephalometric analysis.
• App-aided (CephNinja) and computerized (Dolphin) methods demonstrate accuracy comparable to manual tracings.