BRAR-anchored multimodal dataset

Accurate assessment of alveolar bone resorption from panoramic radiographs is crucial for diagnosing and managing periodontal disease. However, the development of robust, automated diagnostic tools is significantly hampered by a scarcity of large-scale, expertly annotated multimodal datasets. To bridge this gap, we present a comprehensive periodontal dataset collected from 1,104 patients. This dataset uniquely integrates demographic information, full-mouth panoramic radiographs, and detailed clinician-annotated clinical indicators at both patient and tooth levels, with a critical focus on the assessment of alveolar bone resorption. Key clinical variables include the number of missing teeth, implants, retained roots, andopposing pairs of functionalteeth. The dataset introduces a novel Bone Resorption Age Ratio (BRAR), which provides a standardized method for assessing the severity of bone resorption. Data annotations were rigorously performed and validated by experienced periodontists, ensuring high data quality. The dataset’s large scale, combined with its rich multimodal structure and rigorously expert-validated annotations, provides a high-quality, foundational resource. It is poised to significantly accelerate the development and benchmarking of AI-driven diagnostic models and facilitate novel, large-scale clinical research in periodontics, contributing to advances in periodontal disease diagnosis, prognosis, and epidemiology.