Related Experiment Video
Updated: Jul 31, 2025
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
904
Refinement of image quality in panoramic radiography using a generative adversarial network.
Hak-Sun Kim1, Eun-Gyu Ha2, Ari Lee1
1Department of Oral and Maxillofacial Radiology, Yonsei University College of Dentistry, Seoul, South Korea.
Dento Maxillo Facial Radiology
|May 2, 2023
Summary
A generative adversarial network (GAN) model can improve degraded panoramic dental X-rays. The GAN model shows potential for enhancing image quality, particularly for blurred images, but requires further development for noisy images.
Area of Science:
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Panoramic radiography is crucial for dental diagnostics.
- Image quality degradation can hinder accurate interpretation.
- Generative Adversarial Networks (GANs) show promise in image restoration.
Purpose of the Study:
- To develop and evaluate a GAN model for improving image quality in panoramic radiography.
- To assess the clinical usefulness of the GAN model for degraded dental X-rays.
Main Methods:
- A Pix2Pix GAN model was trained on 100 original and 400 degraded panoramic radiographs.
- Degradation methods included blur, noise, and combined effects.
- Model performance was evaluated using peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and radiologist scoring.
Main Results:
- The GAN model significantly improved image quality, with notable enhancement in blurred anterior teeth regions.
- Quantitative metrics (PSNR, SSIM) and qualitative radiologist scores showed improvement.
- The model was less effective on images with combined blur and noise.
Conclusions:
- The developed GAN model demonstrates potential for quantitatively and qualitatively improving degraded panoramic radiographs.
- Further research is needed to optimize the GAN model's performance on noisy radiographic images.
