Related Experiment Video
Updated: Jun 17, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.7K
A real-time interactive restoration system for intraoral digital videos using segment anything model
Yongjia Wu1, Li Zeng1, Yaya Hong2
1Department of Orthodontics, Stomatology Hospital, School of Stomatology, Zhejiang University School of Medicine, Hangzhou, Zhejiang, China.
Digital Health
|August 7, 2024
Summary
This study introduces a real-time interactive restoration system to improve low-quality intraoral videos. The system utilizes the Segment Anything Model for enhanced clinical diagnosis and patient care.
Area of Science:
- Biomedical Imaging
- Computer Vision
- Medical Technology
Background:
- Intraoral environments often yield poor-quality photos and videos, impeding accurate clinical diagnosis.
- Digital dental records require high fidelity for effective analysis and treatment planning.
Purpose of the Study:
- To develop a real-time interactive system for restoring low-quality intraoral videos.
- To enhance the diagnostic value of digital intraoral records.
Main Methods:
- The system uses the Segment Anything Model (SAM) for interactive segmentation of intraoral videos.
- It incorporates real-time intraframe restoration and video enhancement modules.
- Ablation studies were performed to validate system design and performance.
Main Results:
- Achieved high segmentation accuracy with a mean intersection-over-union of 0.977.
- Demonstrated reliable video restoration with PSNR of 37.09 and SSIM of 0.961.
- Clinical applicability was validated by expert evaluation.
Conclusions:
- The interactive restoration system offers a reliable and controllable solution for improving intraoral videos.
- This technology has the potential to significantly benefit both patients and dental professionals.
