AI-Assisted Detection of Periodontal Osseous Defects in Radiographic Images

Periodontal osseous defects are key indicators of periodontal disease severity, requiring precise and early detection to facilitate effective clinical intervention. This study explores an AI-assisted framework for the automated detection of periodontal bone defects using YOLOv8, a state-of-the-art deep learning model optimized for object detection. The proposed approach leverages a dataset of annotated intraoral radiographs, with expert-labeled lesions corresponding to four clinically relevant categories: 1, 2+, crater, and furca. The YOLOv8 model was trained following a structured deep learning pipeline, incorporating advanced augmentation techniques, customized training configurations, and comprehensive evaluation metrics to assess detection accuracy. The results highlight the potential of deep learning in periodontal diagnostics, while also addressing critical challenges such as dataset limitations, class imbalance, and the complexity of small lesion detection. Future research directions focus on expanding dataset diversity, optimizing network architectures, and integrating AI-assisted diagnostics into clinical workflows to enhance real-world applicability.