Slice Extraction and Tilt Correction Forsynthesis of Panoramic X-Rays from CT Image

Panoramic X-ray imaging is a standard diagnostic modality in dentistry, offering a comprehensive view of dental structures. Building AI models for dental diagnostics requires diverse datasets, which are challenging to obtain due to high imaging costs and the need for patient privacy protection. To address this challenge, we propose a method for synthesizing panoramic X-ray images from CT images using an elliptical dental arch and a dynamic rotation center. The method also introduces slice extraction and tilt correction relative to the axial reference plane, which have not been considered in previous approaches. Experiments using CT images from 321 individuals showed that the proposed method achieved a mean Structural Similarity Index (SSIM) of 0.644, outperforming conventional techniques. These results suggest that our method can generate panoramic X-rays similar to actual images and support data augmentation in machine learning.