Diagnostic accuracy of artificial intelligence tools in evaluating the inferior alveolar nerve-third molar relationship: a systematic review.

OBJECTIVE

To determine the diagnostic capability of artificial intelligence (AI) for detecting the relationship between the inferior alveolar nerve (IAN) and the third molar.

METHODS

The search was conducted in 9 electronic databases (PubMed/MEDLINE, EMBASE, LILACS, Web of Science, Scopus, LIVIVO, Computers & Applied Sciences, ACM Digital Library, and Compendex) and gray literature (Google Scholar and ProQuest) for studies published up to April 2025.

RESULTS

Seventeen diagnostic studies were included. The total sample size comprised approximately 22 211 third molars and 16 412 imaging exams. Fourteen studies used 2D images (panoramic radiographs) to diagnose the relationship between third molars and the IAN. Three studies used cone-beam CT (CBCT). The results of the selected studies showed significant heterogeneity in models, imaging examinations, and classification methods for the relationship between the IAN and third molars, as well as divergences in the presentation of metrics.

CONCLUSIONS

Artificial intelligence tools have demonstrated relatively high accuracy in diagnosing the relationship between the IAN and third molars, offering time savings and providing precise, reliable diagnoses for radiologists, oral and maxillofacial surgeons, and general practitioners alike. Although the data accuracy of CBCT models seemed promisingly high, the data were limited, and the level of evidence certainty was low.

ADVANCES IN KNOWLEDGE: Artificial intelligence demonstrates high accuracy in assessing the relationship between the inferior alveolar nerve and third molars, offering efficient and reliable diagnostic support. While CBCT-based AI models show promising results, the current evidence is limited, requiring further research for clinical validation.