Artificial Intelligence for Assessing the Relationship Between Mandibular Third Molars and the Inferior Alveolar Nerve on Panoramic Radiographs: A Systematic Review.

In 30 seconds

This systematic review assessed artificial intelligence (AI) models that use panoramic radiographs to evaluate the relationship between mandibular third molars and the inferior alveolar nerve, with cone-beam computed tomography or expert assessment as reference standards. Eleven studies involving more than 9,700 patients were included. Reported model accuracy ranged from 0.630 to 0.896, with sensitivity and specificity reaching up to 0.92 and 0.95, respectively. No architecture was consistently superior; in studies with human comparisons, AI outperformed less experienced observers. The authors consider AI promising for decision support but call for stronger validation before widespread clinical use.

Key findings

• Across eleven studies with more than 9,700 patients, AI accuracy ranged from 0.630 to 0.896.
• Sensitivity and specificity reached up to 0.92 and 0.95, respectively.
• AI models outperformed less experienced human observers in studies that included this comparison.
• No architecture was consistently superior; validation against CBCT usually produced more reliable metrics than validation based solely on expert opinion.

Why it matters

Panoramic radiographs are commonly used before mandibular third-molar extraction, yet the review notes their limitations in assessing the nerve relationship. AI may help interpret these images and support risk assessment. The variable performance and validation methods across studies mean that reported accuracy should not be assumed to translate uniformly across settings.

What to keep in mind

The review reports variation in datasets, reference standards, and validation methods, preventing identification of a consistently superior architecture. It calls for larger, multicenter, prospective studies with external validation.