Artificial intelligence: is it more accurate than endodontists in root canal therapy?

Abstract

Data sources: The following databases were electronically searched (up to 20 March 2022): PubMed, Scopus, Google Scholar, and Cochrane Library. This was followed by hand-searching the reference lists of the included articles. The search was restricted to articles published in English. The aim of this study was to evaluate the effectiveness of artificial intelligence in identifying, analyzing, and interpreting radiographic features related to endodontic therapy.

Study selection: The selection criteria were limited to trials evaluating the effectiveness of artificial intelligence in identifying, analyzing, and interpreting radiographic features related to endodontic therapy.

Types of studies: Clinical, ex-vivo, and in-vitro trials.

Types of radiographic images: Two-dimensional intra-oral imaging (bitewings and/or periapicals), panoramic radiographs (PRs), and cone beam computed tomography (CBCT).

Exclusion criteria: 1) Case reports, letters, and commentaries; 2) Reviews, conferences, and books; 3) Inaccessible reports.

Data extraction and synthesis: The titles and abstracts of the results of the searches were screened by two authors against the inclusion criteria. The full text of any potentially relevant abstract and title were retrieved for more comprehensive assessment. The risk of bias was assessed initially by two examiners and then by two authors. Any discrepancies were resolved through discussion and consensus.

Results: Out of the 1131 articles which were identified in the initial search, 30 were considered relevant, and only 24 articles were eventually included. The exclusion of the six articles was related to the absence of appropriate clinical or radiological data. Meta-analysis was not performed due to high heterogeneity. Various degrees of bias were detected in more than 58% of the included studies.

Conclusions: Although most of the included studies were biased, the authors concluded that the use of artificial intelligence can be an effective alternative in identifying, analyzing and interpreting radiographic features related to root canal therapy.