Related Experiment Video
Updated: Jan 7, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
A deep learning based automated maxillary sinus segmentation and bone grafts analysis in CBCT images
Fan Yang1, Xing Wu1,2, Yukang Zhang1
1Center for Plastic and Reconstructive Surgery, Department of Stomatology, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, Zhejiang, China.
NPJ Digital Medicine
|December 30, 2025
Summary
A new deep learning system, SA-ai, automates bone gain evaluation after maxillary sinus augmentation. This technology significantly improves workflow efficiency and provides objective, precise measurements for optimizing dental implant therapy.
Area of Science:
- Biomedical Engineering
- Radiology
- Dental Implantology
Background:
- Accurate assessment of bone gain following maxillary sinus augmentation is crucial for successful dental implant outcomes.
- Current manual measurement techniques are time-consuming and prone to variability.
- Objective and efficient methods are needed for longitudinal monitoring of bone graft volume.
Purpose of the Study:
- To validate a fully automated deep learning system (SA-ai) for quantifying bone augmentation after maxillary sinus augmentation.
- To compare the accuracy and efficiency of SA-ai against manual measurement methods.
- To assess the potential of SA-ai in standardizing the clinical evaluation of post-augmentation bone dynamics.
Main Methods:
- A deep learning system integrating 2D U-Net and 3D V-Net was developed for sinus and maxilla segmentation.
- The system was trained and tested on a paired CBCT dataset from 85 patients.
- SA-ai performance was evaluated using Dice coefficient and registration RMSE, with clinical validation against manual measurements.
Main Results:
- The SA-ai system achieved a high Dice coefficient (93.2%) and low registration RMSE (1.046 mm).
- Excellent agreement was observed between SA-ai and manual measurements for bone volume (ICC = 0.993) and other parameters.
- Workflow efficiency was improved over 20-fold compared to manual methods, with confirmed measurement stability.
Conclusions:
- The automated SA-ai system offers an objective and efficient solution for quantifying bone gain after maxillary sinus augmentation.
- This technology has the potential to standardize the clinical evaluation of bone graft volume, including in one-stage implant cases.
- SA-ai facilitates longitudinal monitoring of bone dynamics, optimizing implant therapy planning and outcomes.
