The use of artificial intelligence in the diagnosis of carious lesions: Systematic review and meta-analysis

BackgroundThe use of Artificial Intelligence (AI) has many applications in the healthcare field. Dental caries is a disease with a prevalence rate of over 50% in Brazil. The diagnosis of caries is usually based on a clinical examination and supplementary tests such as X-rays. The accuracy of a diagnostic test is evaluated by its sensitivity, specificity, and accuracy. Various algorithms and neural network configurations are being used for caries diagnosis. ObjectiveThis systematic review evaluated the sensitivity, specificity, and accuracy of using deep machine learning through a convolutional neural network in diagnosing dental caries. MethodsThis systematic review was conducted in accordance with the Preferred Reporting Items for Systematic review and Meta-Analyses (PRISMA) 2020 guidelines and registered with Prospero (ID CRD42024411477). We used the PubMed, MEDLINE, and LILACS databases and MeSH and DECs descriptors in the search. ResultsAfter analyzing the eligibility of the articles, we selected 33 for full-text reading and included 13 in the meta-analysis. We used the sensitivity, specificity, accuracy data, and the number of positive and negative tests to generate a 2x2 table with TP, FP, FN, TN rates, and accuracy. We evaluated the heterogeneity of the SROC curve using the Zhou & Dendurkuri I 2 approach. The results showed that the sensitivity and specificity of the machine learning for detecting dental caries were 0.79 and 0.87, respectively, and the AUC of the SROC curve was 0.885. ConclusionThe literature presented a variety of convolutional neural networks [CNN] architecture, image acquisition methods, and training volumes, which could lead to heterogeneity. However, the accuracy of using artificial intelligence for caries diagnosis was high, making it an essential tool for dentistry.