Medical imaging methods are crucial in dental patient care for diagnosing pathologies related to teeth and surrounding structures. Radiological techniques, such as cone-beam computed tomography (CBCT), are vital for orthodontic diagnosis, treatment planning, and monitoring. With the increasing number of radiological examinations, there is a growing need for comprehensive diagnostic tools. To address this, Artificial Intelligence (AI)-based systems have emerged. Machine Learning (ML) has significantly impacted medical specialties, including orthodontics, offering promising results. ML enables early screening, accurate diagnosis, appropriate treatment, and prediction of treatment-associated toxicity for maxillofacial cysts and tumors. Additionally, ML models are valuable for planning, evaluating, and improving dental implants. The purpose of this work was to develop an algorithm for dentistry aimed at recognizing pathologies observable in second-level instrumental investigations, to present them in a simplified way both to expert dentists and those who intend to approach image reading. Six patients (3M, 3F, ages 29-61) with various dental treatments were selected. UNet was used for segmentation, performing pixel-wise classification to localize and distinguish the edges of structures in the images. The model's architecture allows the input and output to share the same dimensions, facilitating accurate delineation of dental structures. The training utilized a dataset of annotated images to enhance the model's capacity to identify and differentiate various dental treatments and conditions. Once training was complete, the neural network was tested on a separate dataset known as the test set, which consisted of data not used during the former phase. This testing phase aimed to evaluate the neural network’s performance in practical scenarios and provide an objective estimate of its segmentation capabilities. UNet effectively demonstrated its ability to precisely identify and highlight structures of interest, including those that are smaller in size, thereby reducing the manual workload of dental operators.