Related Experiment Video
Updated: Jan 10, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
An innovative AI-based dual segmentation application for head surgery
1Department of Oral and Craniomaxillofacial Surgery and 3D Print Lab, University Hospital Basel, Basel, Switzerland; Medical Additive Manufacturing Research Group (Swiss MAM), Department of Biomedical Engineering, University of Basel, Allschwil, Switzerland.
International Journal of Oral and Maxillofacial Surgery
|November 26, 2025
Summary
This study introduces a dual-model artificial intelligence (AI) system for automated craniofacial segmentation in CT scans. The AI system significantly reduces manual segmentation time while maintaining clinical accuracy for surgical planning.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
- Computational anatomy
Background:
- Accurate anatomical segmentation in CT imaging is crucial for head and neck surgery diagnostics and surgical planning.
- Manual segmentation is labor-intensive, time-consuming, and prone to inconsistencies.
Purpose of the Study:
- To develop and validate a dual-model AI system for automated segmentation of craniofacial structures in CT scans.
- To assess the accuracy and robustness of the AI system compared to manual segmentation.
Main Methods:
- A two-stage nnU-Net-based AI approach was employed, utilizing a coarse global model and a fine local model.
- 388 clinical CT scans were processed, with segmentations evaluated using Dice Similarity Coefficient (DSC), Mean Surface Distance (MSD), and Hausdorff Distance (HD).
Main Results:
- The AI system achieved high accuracy, with mean DSC scores of 0.963 for mandible/skull and 0.986 for soft tissue.
- Low Mean Surface Distance values were observed for maxillary sinus (0.134 mm) and mandible (0.150 mm).
- The system demonstrated robustness across diverse imaging protocols, including low-resolution scans.
Conclusions:
- The dual-model AI system automates craniofacial segmentation efficiently and accurately, reducing manual effort.
- This open-access framework provides a scalable solution for diagnostic, surgical, and educational applications in craniofacial care.
- The AI system ensures clinical precision, supporting improved patient outcomes in head and neck surgery.
