In vitro Assessment of a Deep Learning-Based System for Computer-Aided Diagnosis and Treatment Planning in...

Area of Science:

• Orthodontics
• Artificial Intelligence
• Medical Imaging

Background:

• Automating orthodontic diagnosis and treatment planning can enhance efficiency and accuracy.
• Digital orthodontic models are increasingly utilized in clinical practice.

Purpose of the Study:

• To evaluate the efficacy of a deep learning-based artificial intelligence (AI) system for computer-aided orthodontic diagnosis and treatment planning.
• To assess the performance of AI in analyzing digital orthodontic models.

Main Methods:

• A deep learning AI system was developed and trained on 100 three-dimensional digital orthodontic models.
• The AI system was tasked with recognizing malocclusions and generating treatment recommendations.
• AI-generated results were compared against evaluations by qualified orthodontists.

Main Results:

• The AI system demonstrated high accuracy in malocclusion categorization.
• Sensitivity and specificity for malocclusion identification exceeded 90% in most cases.
• AI performance was comparable to that of experienced orthodontists.

Conclusions:

• The findings support the potential of AI in automating orthodontic diagnosis and treatment planning.
• Further clinical validation is necessary to confirm the efficacy of AI in orthodontics.
• AI holds promise for improving efficiency and accuracy in orthodontic workflows.