Automated Skeletal Classification with Lateral Cephalometry Based on Artificial Intelligence.

Area of Science:

• Orthodontics
• Medical Imaging
• Artificial Intelligence

Background:

• Lateral cephalometry is standard for orthodontic diagnosis but is time-consuming and prone to errors.
• Manual landmark tracing in conventional methods introduces inter- and intra-observer variability.

Purpose of the Study:

• To develop an accurate and robust skeletal diagnostic system using a convolutional neural network (CNN).
• To create a one-step, end-to-end diagnostic system for lateral cephalograms.

Main Methods:

• A multimodal CNN model was trained on 5,890 lateral cephalograms and demographic data.
• Transfer learning and data augmentation techniques were employed for model optimization.
• Diagnostic performance was assessed using statistical analysis and receiver operating characteristic curves.

Main Results:

• The CNN system demonstrated over 90% sensitivity, specificity, and accuracy for vertical and sagittal skeletal diagnosis.
• The highest accuracy for vertical classification reached 96.40%.
• The model achieved an average area under the curve greater than 95%, indicating excellent performance.

Conclusions:

• The proposed CNN-incorporated system offers a potential advancement for skeletal orthodontic diagnosis.
• This AI-driven approach eliminates the need for complex intermediary diagnostic steps, enhancing efficiency and accuracy.