Evaluation of orthognathic surgery planning with artificial intelligence: a prospective, comparative study

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Cephalometric analysis is crucial for orthognathic surgery planning.
• Traditional methods rely on manual measurements, which can be time-consuming and prone to error.
• Deep learning-based AI offers potential for automated cephalometric analysis.

Purpose of the Study:

• To evaluate the accuracy of AI-driven cephalometric analyses.
• To compare AI measurements against a 3D CT scan gold standard.
• To assess the reliability of AI tools (NemoCeph 2D, OrthoDx, AudaxCeph, WebCeph) in orthognathic surgery patients.

Main Methods:

• 3D CT scans were obtained from orthognathic surgery candidates.
• 3D cephalometric software established gold-standard landmark positions.
• AI programs automatically identified landmarks on 2D cephalometric images derived from 3D scans.
• Statistical analysis compared AI measurements with the 3D gold standard.

Main Results:

• The ANB angle showed no significant difference (p=0.061).
• Significant discrepancies (p<0.05) were found for SNA, SNB, Wits appraisal, Y Axis Angle, and facial height ratios.
• AI analyses demonstrated notable errors when compared to the 3D CT gold standard.

Conclusions:

• Current deep learning AI cephalometric analyses exhibit inaccuracies.
• These AI algorithms require further development for reliable use in orthognathic surgery planning.
• Clinical implementation of AI in this field necessitates algorithmic refinement.