Use of Artificial Intelligence in Diagnosing Vertical Root Fractures-A Systematic Review

Area of Science:

• Dentistry
• Radiology
• Artificial Intelligence

Background:

• Vertical root fractures (VRFs) are difficult to diagnose due to subtle radiographic signs.
• Existing imaging techniques and AI performance across modalities are not well-established.

Purpose of the Study:

• To systematically review and compare AI-assisted VRF detection across periapical radiography, panoramic radiography, and CBCT.
• To evaluate diagnostic performance, methodological quality, and limitations of AI in VRF detection.

Main Methods:

• Systematic literature review up to January 2025 across major databases.
• Included studies used AI for VRF detection in periapical, panoramic, or CBCT images.
• Data extraction focused on AI models, datasets, imaging, validation, and diagnostic metrics; risk of bias assessed with QUADAS-2.

Main Results:

• Ten studies utilized predominantly CNN-based AI models.
• CBCT-based AI achieved highest accuracy (91.4-97.8%) and specificity (90.7-100%).
• Periapical radiography models showed high accuracy (up to 95.7%), while panoramic radiography models had lower sensitivity but high precision.

Conclusions:

• AI-assisted VRF detection is promising, especially with CBCT.
• Current evidence is limited by methodological heterogeneity and insufficient clinical validation.
• Further research with standardized methodologies and clinical validation is required.