Osteoporosis detection using artificial intelligence

Objectives: To evaluate the effectiveness of Vision Transformer (ViT) architecture for osteoporosis detection using dental panoramic radiographs. Methods: Digital panoramic radiographs of female patients across multiple age groups were analyzed. Mandibular indices including Mental Index, Panoramic Mandibular Index, Gonial Index, Antegonial Index, and Antegonial Depth were measured by two oral radiologists. A ViT-based deep learning model was trained and evaluated for classification performance. Results: The ViT model achieved an accuracy of 98.33%, sensitivity of 96.50%, specificity of 98.90%, and an F1-score of 97.68%, outperforming conventional convolutional neural network models. Conclusions: Transformer-based deep learning models demonstrate high diagnostic potential for osteoporosis screening using dental radiographs and may support early detection and preventive intervention strategies.