LRVRG: a local region-based variational region growing algorithm for fast mandible segmentation from CBCT images

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• 3D Reconstruction

Background:

• Accurate segmentation of the mandible from cone-beam computed tomography (CBCT) images is crucial for dental research and clinical diagnosis.
• Existing methods may struggle with the fuzzy boundaries often present in CBCT data.

Purpose of the Study:

• To develop an efficient and accurate algorithm for segmenting the mandible from CBCT images.
• To create high-fidelity 3D mandible models for subsequent analysis and diagnosis.

Main Methods:

• A local region-based variational region growing (LRVRG) algorithm integrating local region and shape prior.
• Iterative energy minimization to identify mandible pixels within CBCT slices.
• Utilizing previous slice segmentation as shape prior for subsequent slices.
• Marching Cubes algorithm for final 3D model reconstruction.

Main Results:

• The LRVRG algorithm successfully generates satisfactory 3D mandible models from CBCT images.
• The method effectively addresses the challenge of fuzzy boundaries in CBCT data.
• Quantitative comparisons show the proposed method achieves state-of-the-art performance in mandible segmentation.

Conclusions:

• The developed LRVRG method is both efficient and accurate for 3D mandible model segmentation from CBCT.
• This approach offers a reliable tool for enhancing dental diagnostics and research.