Artificial Intelligence and 3D Imaging in Orthodontics: Predictive Analysis of Soft Tissue Changes and Treatment Outcomes

Abstract

Background: The integration of artificial intelligence (AI) and three-dimensional (3D) imaging technologies has significantly advanced orthodontic diagnostics and treatment planning. This study evaluates the predictive accuracy of AI models in forecasting soft tissue changes and treatment outcomes using 3D imaging data, aiming to enhance treatment precision.

Methods: An in vitro experimental study was conducted using 3D-printed orthodontic models based on anonymized cone-beam computed tomography (CBCT) scans. The study involved the fabrication of orthodontic appliances and the application of simulated orthodontic forces using a Tensometer 5000 Universal Testing Machine.  AI algorithms, including deep learning models, were trained on pre-treatment 3D images and treatment plans to predict post-treatment soft tissue outcomes. The predictive models accounted for tooth movement, facial growth, and soft tissue response. The study, conducted from January 3, 2024, to May 17, 2024, involved generating high-resolution orthodontic models, applying simulated orthodontic

Results: AI models demonstrated high predictive accuracy, with the DeepConvNet model achieving a mean absolute error (MAE) of 0.42 mm and a root mean square error (RMSE) of 0.53 mm. The correlation coefficient between predicted and actual post-treatment outcomes indicated a strong positive relationship. Soft tissue changes averaged 0.30 mm across key facial regions. The Activator appliance resulted in the highest mean change of 0.35 mm, while force application showed a linear relationship with displacement, where higher forces produced greater tissue movement. The Force Sensor Pro exhibited superior accuracy and precision compared to the Tensometer 5000.

Conclusions: The study highlights the potential of AI and 3D imaging technologies to improve the prediction of soft tissue changes and treatment outcomes in orthodontics. The DeepConvNet model provided the most accurate predictions, and the Activator appliance showed the greatest efficacy in inducing soft tissue changes. These findings suggest that AI-driven predictive models and advanced imaging can lead to more precise and individualized orthodontic treatments, enhancing patient satisfaction and clinical outcomes. Further research with larger datasets and clinical trials is recommended to validate and refine these models.