Related Experiment Video
Updated: Jan 5, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
A review on multiplatform evaluations of semi-automatic open-source based image segmentation for cranio-maxillofacial
Jürgen Wallner1, Michael Schwaiger1, Kerstin Hochegger2
1Medical University of Graz, Department of Oral and Maxillofacial Surgery, Auenbruggerplatz 5/1, Graz 8036, Austria; Computer Algorithms for Medicine Laboratory, Graz 8010, Austria.
Computer Methods and Programs in Biomedicine
|October 15, 2019
Summary
Open-source segmentation methods like GrowCut and Canny offer high-quality, semi-automatic image segmentation for cranio-maxillofacial surgery. These license-free tools provide viable clinical alternatives for surgical planning and treatment visualization.
Area of Science:
- Medical Imaging
- Computer-Assisted Surgery
- Biomedical Engineering
Background:
- Image-based segmentation is crucial for cranio-maxillofacial surgery diagnosis and treatment.
- Clinical adoption of segmentation software is hindered by resource constraints.
- A need exists for evaluating open-source segmentation solutions.
Purpose of the Study:
- To assess and review the segmentation quality of license-free methods.
- To evaluate the clinical utility of open-source segmentation on multiple platforms.
- To identify effective segmentation algorithms for cranio-maxillofacial applications.
Main Methods:
- Evaluated open-source segmentation algorithms (GrowCut, Canny, etc.) on 3D Slicer, MITK, and MeVisLab using CT data.
- Compared algorithm performance against expert-defined ground truth segmentations (n=20).
- Assessed segmentation accuracy using Dice Score Coefficient (DSC), Hausdorff Distance (HD), and Pearsons correlation coefficient (r).
Main Results:
- GrowCut (DSC 85.6%) and Canny (DSC 82.1%) demonstrated the highest segmentation accuracy.
- No significant statistical differences were found between assessed parameters (p < 0.05).
- High correlation coefficients (r > 0.94) and time-saving segmentations were observed.
Conclusions:
- GrowCut and Canny provide high-quality, semi-automatic segmentation for cranio-maxillofacial applications.
- These open-source methods offer practical, cost-effective alternatives for clinical use.
- This study presents a reproducible, multi-platform evaluation of license-free segmentation for patient-individualized medicine.
