Clinically applicable artificial intelligence system for dental diagnosis with CBCT

Area of Science:

• Dentistry
• Artificial Intelligence
• Medical Imaging

Background:

• Cone Beam Computed Tomography (CBCT) is crucial for dental diagnostics.
• Accurate interpretation of CBCT scans requires specialized expertise.
• AI offers potential to enhance diagnostic capabilities in dentistry.

Purpose of the Study:

• To evaluate the real-time performance and clinical effectiveness of a novel AI system for CBCT imaging diagnosis.
• To assess the impact of the AI system on diagnostic accuracy when used by dentists.
• To determine the safety and clinical utility of the AI system in a dental setting.

Main Methods:

• A deep learning-based AI system with five modules (ROI-localization, tooth-localization/numeration, periodontitis, caries, periapical-lesion localization) was developed.
• The AI system was trained on 1346 CBCT scans using Convolutional Neural Networks (CNNs).
• A clinical evaluation involved 24 dentists examining 30 CBCT scans, with one group aided by the AI system and the other unaided.

Main Results:

• The AI-assisted group showed higher overall sensitivity (0.8537) compared to the unaided group (0.7672).
• Specificity was also higher in the AI-assisted group (0.9672) versus the unaided group (0.9616).
• The difference in diagnostic performance between the aided and unaided groups was statistically significant (p=0.032).

Conclusions:

• The novel AI system significantly enhances dentists' diagnostic capabilities in CBCT imaging.
• The AI system demonstrates real-time performance, clinical effectiveness, and safety.
• AI integration in dentistry holds promise for improving diagnostic accuracy and patient care.