PointIpg: an AI Model for Assessing Lower Incisor-Pg as a Cephalometric Parameter for the Anterior Limit of Dentition

Integrating artificial intelligence into healthcare has been revolutionary, enhancing diagnostic, prognostic, and therapeutic applications. Artificial intelligence has significantly improved accuracy and efficiency in medical imaging, leading to better patient outcomes. In dentistry, particularly in orthodontics, artificial intelligence is essential for automating image-based diagnosis and treatment planning, allowing specialists to evaluate anatomical structures with increased precision. This study suggests an artificial intelligence-based approach to determine the Lower Incisor—Pg, a cephalometric parameter that evaluates the sagittal position of the lower incisor, which is an essential aesthetic and functional factor in orthodontics, affecting facial harmony. Traditional manual methods for identifying this parameter on lateral cephalometric radiographs could be subjected to variability and human error. The proposed artificial intelligence model, called PointIpg, employs deep learning algorithms to automatically identify the Lower Incisor—Pg parameter and assess lower incisor’s potential movement within the lower dental arch, offering orthodontists a tool for accurate treatment planning.PointIpg was trained using a dataset of 996 teleradiographs, from which four cephalometric landmarks (Po, Or, B1, Pg) were extracted using Delta-Dent software and a PointNet++ architecture. The model achieved a Mean Absolute Error of 0.03 mm and an overall accuracy of 97.5%, enhancing precision and consistency in orthodontic planning as a reliable alternative to manual methods.