Classification of Maxillofacial Morphology by Artificial Intelligence Using Cephalometric Analysis Measurements

Area of Science:

• Orthodontics
• Artificial Intelligence
• Cephalometric Analysis

Background:

• Maxillofacial morphology is crucial for orthodontic diagnosis and treatment planning.
• Sassouni's classification exists, but lacks standardization in patient assignment.
• Accurate classification is needed to standardize treatment approaches.

Purpose of the Study:

• To develop an artificial intelligence (AI) model for accurate maxillofacial morphology classification.
• To standardize the classification of maxillofacial morphology using cephalometric data.
• To improve orthodontic diagnosis and treatment planning through AI.

Main Methods:

• Utilized cephalograms from 220 adult patients (≥18 years).
• Employed a random forest classifier with eight cephalometric measurements and gender as input features.
• Trained and tested models using k-fold cross-validation for horizontal, vertical, and combined classifications.

Main Results:

• Achieved high classification accuracies: 0.823 for combined, 0.986 for anteroposterior, and 0.850 for vertical.
• The ANB angle was identified as the most important feature (0.3519).
• The AI model demonstrated robust performance in classifying maxillofacial morphology.

Conclusions:

• The developed AI model accurately classifies maxillofacial morphology.
• This AI approach offers a standardized method for classification in orthodontics.
• Further improvements are possible with increased data for the AI model.