Segmentation and reconstruction of vascular structures for 3D real-time simulation

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Computational Anatomy

Background:

• Accurate 3D modeling of patient-specific vascular structures is crucial for medical simulations.
• Existing methods often struggle with centerline accuracy, cross-section fitting, and topological preservation in complex vascular networks.

Purpose of the Study:

• To develop and validate a robust two-step technique for generating smooth and accurate surfaces of patient-specific vascular structures.
• To improve upon existing meshing methods for enhanced topological accuracy and reduced artifacts in vascular modeling.

Main Methods:

• Vessel segmentation using a level set method followed by homotopic thinning to obtain initial centerlines.
• An iterative ellipse fitting scheme to refine centerline positions and cross-section orientations.
• Topologically preserved quadrilateral patch surface reconstruction, improving upon Felkel's method by handling multiple parent/child vessels and reducing undersampling.

Main Results:

• The proposed technique achieves accurate and smooth vascular centerlines with improved cross-section fitting and orientation.
• Surface reconstruction demonstrates topological preservation, reduced artifacts, and adaptability to varying cross-section distributions.
• Experimental results on phantom and real datasets show a favorable balance between surface smoothness, triangle count, and distance error.

Conclusions:

• The developed two-step method effectively generates accurate and smooth 3D models of vascular structures.
• This technique offers significant improvements over existing methods, particularly for complex and branching vessels.
• The approach is suitable for applications in interventional radiology simulations, virtual endoscopy, and general 3D model reconstruction.