Related Experiment Video
Updated: Aug 6, 2026
06:54
Photorealistic Learned Landscapes for Augmented Reality
Published on: June 27, 2025
Controllable Panoramic Radiograph Synthesis Using a Generative Model
Y Y Lin1,2, W T Fu1,2, X L Guo1,3
1State Key Laboratory of Oral & Maxillofacial Reconstruction and Regeneration, Key Laboratory of Oral Biomedicine Ministry of Education, Hubei Key Laboratory of Stomatology, School & Hospital of Stomatology, Wuhan University, Wuhan, Hubei, China.
Journal of Dental Research
|July 18, 2026
Summary
A new AI model, PRGen, generates realistic panoramic radiographs (PRs) with annotations. This addresses data scarcity and improves AI diagnostic accuracy in dentistry.
Area of Science:
- Dental imaging
- Artificial intelligence
- Medical image generation
Background:
- Panoramic radiography (PR) is a key dental diagnostic tool.
- Automated interpretation of PRs is crucial for efficiency and consistency.
- AI development for PR faces challenges like data scarcity and annotation imbalance.
Purpose of the Study:
- To develop PRGen, a novel generative model for panoramic radiographs.
- To overcome limitations of data scarcity, privacy, and annotation imbalance in AI-based PR interpretation.
- To synthesize realistic PRs and paired masks from text descriptions and sketches.
Main Methods:
- Developed PRGen using 50,127 paired text-image samples.
- Enabled control of dental anatomy via text and sketches.
- Synthesized realistic PRs and segmentation masks.
- Evaluated generated images using internal/public datasets and radiologist assessments.
Main Results:
- Incorporating PRGen-synthesized images improved segmentation Dice score by 47.59% and AUC by 11.53%.
- Over 82% of synthesized PRs were deemed clinically realistic by radiologists.
- External validation showed an average Dice improvement of 25.58% for segmentation.
Conclusions:
- PRGen effectively generates high-quality, mask-annotated PRs.
- The model reduces reliance on manual annotations, aiding AI development.
- PRGen facilitates more reliable automated analysis and clinical translation of AI in dentistry.
