A marcação cefalométrica automática com inteligência artificial é confiável? revisão sistemática com metanálise das medidas cefalométricas SNA, SNB, ANB E WITS

• Autor(a) principal: Mota, Isadora Maria Batista da Silva
• Data de Publicação: 2022
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Institucional da UFS
• Texto Completo: http://ri.ufs.br/jspui/handle/riufs/17213
• topic: Cefalometria; Inteligência artificial; Software; Ortodontia; Cephalometrics; Artificial intelligence; Software; Orthodontics; CIENCIAS DA SAUDE::ODONTOLOGIA

Introduction: Cephalometric analysis stands out for its importance in dentistry, specifically in orthodontics, for the diagnosis and treatment of malocclusion and skeletal discrepancies. Recently, manual tracing performed by specialists has been gradually replaced by software that uses artificial intelligence (AI). Thus, it becomes relevant to assess the accuracy of automatic systems to perform this task. Objective: To evaluate, through a systematic review (SR) with meta-analysis, the reliability of the use of artificial intelligence for cephalometric marking. Materials and methods: In order to answer the question “Is automatic cephalometric marking with artificial intelligence reliable?” A search was carried out on the search platforms PubMed, Sccopus, Embase, Web of Science, LILACS and Google Scholar, using the Population, Intervention, Comparison, Outcome and Study Design (PICOS) search strategy. Treatment effects were defined as standardized mean difference (SMD) and 95% confidence intervals (CI) were established. To assess the risk of bias, the Joanna Briggs questionnaire was used for non-randomized studies. The GRADE tool was used to assess the quality of evidence from the systematic review. Results: The search strategy used identified a total of 1041 articles. Of these, 32 were selected in full text, and 14 were included in the systematic review after careful analysis. Of these 14, 5 studies were included in the meta-analysis to assess macro cephalometric SNA, SNB, ANB and Wits. In this analysis, we obtained Tau2=0.04, Chi2=31.70, and p value=0.01, indicating the presence of statistical heterogeneity, I2=50%, revealing moderate heterogeneity. The SMD presented a total value of -0.05, with a confidence interval of -0.19 to 0.09, indicating that there is no statistical difference between the automatic measurement performed by AI and the manual measurement, however, with a size of small effect. Conclusion: We obtained positive and promising results in most studies about the accuracy in identification of cephalometric landmarks. Although strength of evidence from meta-analysis was considered very low, included studies had a low risk of bias.