Automatic segmentation of 3D micro-CT coronary vascular images

Area of Science:

• Biomedical Imaging
• Computational Biology
• Medical Image Analysis

Background:

• Existing 3D angiographic image reconstruction methods often focus on limited vascular segments.
• Accurate, large-scale vascular models are crucial for whole-organ simulations in coronary hemodynamics, autoregulation, and tissue oxygen delivery.

Purpose of the Study:

• To develop an automated and sub-voxel accurate algorithm for reconstructing entire vascular networks.
• To meet the demand for high-fidelity vascular meshes in advanced physiological simulations.

Main Methods:

• A novel vascular reconstruction algorithm combining automatic vessel seeding and tracking.
• Radius detection utilizing active contours for enhanced precision.
• Validation on synthetic data for topological and morphological accuracy.
• Application and validation on micro-CT scanned rat coronary vasculature.

Main Results:

• Synthetic data testing showed sub-voxel errors (<0.5 voxels) for centerline and radius detection.
• Initial seed direction accuracy was within 3 degrees.
• Real-world data application on rat coronary vasculature demonstrated <10% radius error for vessels >2 voxels in radius.

Conclusions:

• The developed algorithm achieves automated processing and sub-voxel accuracy for 3D vascular network reconstruction.
• This method provides anatomically accurate vascular meshes suitable for whole-organ simulations.
• The findings support advancements in computational hemodynamics and oxygen delivery research.