Is automatic cephalometric software using artificial intelligence better than orthodontist experts in landmark...

Area of Science:

• Radiology
• Artificial Intelligence in Medicine
• Orthodontics

Background:

• Cephalometric analysis is crucial in orthodontics and other dental fields.
• Manual digitization of cephalograms is time-consuming and prone to variability.
• Evaluating automated digitization techniques is essential for improving workflow efficiency and accuracy.

Purpose of the Study:

• To assess the accuracy and efficiency of artificial intelligence (AI) algorithms for automatic cephalogram digitization.
• To compare the performance of different AI-based machine learning programs in landmark identification.
• To determine the success rate of AI in localizing hard and soft tissue cephalometric points.

Main Methods:

• Lateral cephalograms of 43 patients were digitized manually and with AI assistance (MyOrthoX, Angelalign, Digident).
• Image J was used to extract coordinates for 32 cephalometric points (11 soft tissue, 21 hard tissue).
• Mean radical errors (MRE) and successful detection rates (SDR) were calculated at 1.0 mm, 1.5 mm, and 2.0 mm thresholds; statistical analysis was performed using ANOVA.

Main Results:

• All three AI methods achieved detection rates exceeding 85% at the 2 mm threshold, considered clinically acceptable.
• The Angelalign system demonstrated a detection rate above 78% even at the stricter 1.0 mm threshold.
• Significant time differences were observed between AI-assisted and manual tracing, highlighting AI's efficiency gains.

Conclusions:

• AI assistance can enhance the efficiency of cephalometric tracings without compromising accuracy.
• Automated digitization using AI shows promise for routine clinical practice and research settings.
• AI algorithms offer a reliable alternative to manual cephalogram analysis.