Inteligência artificial e cirurgia de terceiro molar, um novo paradigma na previsão e prevenção da parestesia: Uma revisão de literatura

Abstract

Third molars play a significant role due to their association with complications arising from their dental inclusion, such as cavities, periodontal disorders, cyst formation, pericoronaritis, and paresthesias. Paresthesia occurs when there is damage to the inferior alveolar nerve during extraction, resulting in a decrease or loss of sensitivity in a specific region. Imaginological evaluation plays a crucial role in diagnosis and planning. Panoramic radiography is commonly used to identify the relationship between the lower third molar and the mandibular canal. However, this technique has limitations as it is a two-dimensional examination. Computed tomography provides more accurate three-dimensional images, but its use is restricted due to radiation dose and relative cost. Faced with these challenges, artificial intelligence emerges as a promising tool in identifying the risk of paresthesia in patients undergoing third molar surgery. This tool includes subareas such as Machine Learning and Deep Learning, presenting the potential to process images, extract relevant features, and assist in diagnosis. Therefore, the aim of this study was to conduct an integrative literature review on the use of artificial intelligence in processing imaging exams to evaluate the proximity relationship between the lower third molar and the mandibular canal. The guiding question was: "In patients with lower third molars, can artificial intelligence help identify the relationship between the lower third molar and the inferior alveolar nerve compared to the assessment by a dentist to predict the risk of paresthesia?" Evidence was searched in the PubMed, BVS, EMBASE, and SCOPUS databases using MeSH descriptors. Observational studies, randomized or controlled clinical trials, and prospective and retrospective studies were considered. Twenty-five articles were identified, of which seven were selected for this study. Among the results, the development of Explainable Artificial Intelligence stands out for detecting the relationship between the lower third molar and the mandibular canal, as well as the application of convolutional neural networks for evaluating panoramic radiographs in detecting intimate contact between these structures. The comparison between AI assessment and professional evaluation demonstrated the AI's ability to match or surpass the accuracy of an experienced dentist. However, more studies are needed to validate the effectiveness of these methods in different populations and clinical contexts.