Carotid Artery Wall Segmentation in Multispectral MRI by Coupled Optimal Surface Graph Cuts

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Carotid artery imaging is crucial for stroke risk assessment.
• Accurate segmentation of the carotid artery bifurcation is challenging.
• Existing segmentation methods often require extensive manual input.

Purpose of the Study:

• To develop and validate a novel 3D segmentation algorithm for the carotid artery wall.
• To improve the accuracy and efficiency of carotid artery bifurcation segmentation from MR images.
• To reduce the manual effort required for carotid artery segmentation.

Main Methods:

• A 3D coupled optimal surface graph-cut algorithm was developed.
• The algorithm integrates inner and outer border segmentation into a single graph cut.
• Cost functions utilize multi-sequence MR information, requiring only three manual seed points.

Main Results:

• Quantitative validation on 57 carotid arteries yielded high Dice overlap: 0.86 ± 0.06 (complete vessel) and 0.89 ± 0.05 (lumen).
• Reproducibility tests on 60 scans showed excellent agreement (ICC 0.96 for lumen volume, 0.74 for complete vessel volume).
• The method demonstrated robustness and good agreement between baseline and follow-up segmentations.

Conclusions:

• The proposed graph-cut algorithm provides accurate and reproducible 3D segmentation of the carotid artery wall from MR images.
• This automated approach significantly reduces manual intervention, enhancing clinical utility.
• The algorithm shows promise for improved carotid artery analysis in stroke risk stratification.