Detection of dental malocclusions according to angle classification using artificial intelligence algorithms developed with deep learning on intraoral orthodontic photographs

Aim: This study aims to evaluate the feasibility of classifying Angle malocclusion using intraoral right and left buccal photographs with a deep learning approach, and to develop a model based on the YOLOv8x-cls architecture. Material and Method: A total of 1296 standardized intraoral images were used. The data were split into training (80%), validation (10%), and test (10%) sets. The model was trained for 300 epochs with an input size of 640 pixels and a batch size of 16. Data augmentation techniques were applied to address class imbalance. Model performance was assessed using accuracy, precision, recall, and F1 score. Results: The overall classification accuracy was 78.2%. The highest performance was observed in Class II malocclusion, with an accuracy of 86.47% and an F1 score of 0.86. For Class I, the accuracy was 80.45% with an F1 score of 0.74, while Class III showed an accuracy of 89.47% but a lower F1 score of 0.63. These findings highlight the impact of data imbalance and the importance of augmentation strategies on model performance. Conclusion: The findings demonstrate that Angle classification based on intraoral images can be effectively performed using AI-assisted systems. The proposed model shows promising potential as a clinical decision support tool. Keywords: Angle classification, Artificial intelligence, Classification, Deep learning, Intraoral photography, Malocclusion, YOLOv8x-cls.