Automatic segmentation of mandibular canal in cone beam CT images using conditional statistical shape model and fast...

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Anatomical Modeling

Background:

• Accurate segmentation of the mandibular canal is crucial for successful dental implant surgery.
• Cone beam CT (CBCT) scans are widely used for dental imaging, but precise canal segmentation remains challenging.
• Existing segmentation methods often struggle with image noise and anatomical variations.

Purpose of the Study:

• To develop and evaluate a novel segmentation method for the mandibular canal in CBCT scans.
• To combine anatomical and statistical information for enhanced segmentation accuracy.
• To improve the reliability of mandibular canal segmentation for computer-guided dental implant surgery.

Main Methods:

• A three-step approach involving low-rank decomposition for preprocessing.
• Training a conditional statistical shape model (SSM) for accurate mandibular bone segmentation.
• Utilizing fast marching with a novel speed function to precisely localize the mandibular canal.

Main Results:

• The method was successfully applied to 120 CBCT datasets.
• The conditional SSM demonstrated strong compactness, specificity, and generalization abilities.
• The framework effectively segmented the mandibular bone and canal, even in noisy scans and cases with mild bone resorption.

Conclusions:

• The proposed method significantly outperforms two other recent techniques in quantitative analysis.
• The framework is robust and suitable for computer-guided dental implant surgery.
• This approach enhances the accuracy and reliability of mandibular canal segmentation in CBCT imaging.