Artificial intelligence for the early detection of periapical lesions: literature review

Abstract

Periapical lesions are described as inflammatory or infectious processes affecting the tissues around the dental apex, usually in response to irritation caused by pulp necrosis and the subsequent spread of microorganisms and their toxins to the periapical area. This study addresses the early detection of periapical lesions, conditions that compromise oral health, through the use of artificial intelligence (AI). Its objective was to explore the use of AI for the early detection of periapical lesions in periapical radiographs, evaluating its accuracy and efficiency compared to traditional diagnostic methods. An exhaustive bibliographic review was conducted in the PubMed, Scopus, and Google Scholar databases, selecting studies published between 2019 and 2024 that compared AI algorithms, mainly convolutional neural networks, with conventional radiographic interpretation methods; specific inclusion and exclusion criteria were applied, and meta-analysis techniques were used to consolidate the results. The findings revealed that AI algorithms achieved an average sensitivity of 92 % and specificity of 89 %, in contrast to traditional methods that present values of 80 % and 75 %, respectively, in addition to reducing diagnosis time by 35 %. In conclusion, the application of AI in the detection of periapical lesions significantly improves diagnostic accuracy and process efficiency, allowing for more timely therapeutic planning and optimizing clinical care, positioning itself as an innovative and promising tool in the field of dentistry.