Segmentação orientada por inteligência artificial da câmara pulpar de incisivos centrais superiores em imagens de tomografia computadorizada de feixe cônico

Abstract

With the growing application of artificial intelligence (AI) in dentistry, it is important to test the automation of segmentation of specific areas to estimate chronological age, in order to provide professionals with faster and more accurate data. Thus, the objective of this study was to test a convolutional neural network (CNN) model for segmentation of the pulp chamber in Brazilian adult individuals, through the analysis of tomographic images of upper central incisors. A total of 662 cone beam computed tomography (CBCT) scans from patients aged between 18 and 60 years were analyzed, totaling 1,324 teeth. The process involved training two neural networks using the coronal and sagittal tomographic slices of the selected teeth: the first (YOLOv8) identified and isolated the teeth in the original images; and the second (YOLOv8) performed the segmentation of the areas of interest (pulp chamber and dental crown). The automated identification and segmentation of the areas of interest (pulp chamber and dental crown) achieved excellent performance metrics, with an accuracy greater than 0.9 for both the pulp chamber and the dental crown. It can be concluded that the CNN applied in this study demonstrated a high level of accuracy for segmenting the pulp chamber in CBCT images of upper central incisors. The automation of segmentation of specific areas for age estimation from tomographic images can become a valuable tool, providing professionals with faster and more accurate information, in addition to contributing to significant advancements in both dentistry and forensic medicine.