Three-Dimensional Blood Vessel Segmentation and Centerline Extraction based on Two-Dimensional Cross-Section...

Area of Science:

• Medical imaging
• Image analysis
• Computational anatomy

Background:

• Accurate segmentation of tubular structures like blood vessels is crucial for medical diagnostics and interventions.
• Existing methods often face challenges with speed and accuracy in volumetric datasets.

Purpose of the Study:

• To develop a novel, semi-automatic method for blood vessel segmentation and centerline extraction.
• To improve the efficiency and accuracy of tubular structure analysis in medical imaging.

Main Methods:

• A semi-automatic approach tracking vessel trees from a seed point.
• Utilizing 2D cross-section analysis with a novel circle enhancement filter (single-scale or multi-scale).
• Optional active contour post-processing for enhanced segmentation.

Main Results:

• Achieved a cross-sectional centerline error below 0.8 pixels.
• Obtained a Dice coefficient of 80% ± 2.7% for segmentation, improving to 94% ± 2.4% with post-processing.
• Demonstrated significant speed improvements (over 18x faster than Frangi vesselness).

Conclusions:

• The proposed 2D cross-section analysis method offers a fast and accurate solution for blood vessel segmentation and centerline extraction.
• The method's efficiency and accuracy make it suitable for various medical applications involving volumetric data analysis.