Vascular decomposition using weighted approximate convex decomposition.

Area of Science:

• Medical imaging analysis
• Computational anatomy
• Vascular modeling

Background:

• Computed tomography (CT) angiography is crucial for stroke treatment, requiring detailed vascular pathology analysis.
• Vascular complexity and variability pose challenges in precisely relating vessel geometry to arterial pathology.
• Accurate decomposition of complex vascular structures is needed for advanced computer-aided diagnosis.

Purpose of the Study:

• To develop a novel convex shape decomposition strategy for analyzing complex vascular structures.
• To synthesize a weighted approximate convex decomposition (WACD) method for vascular decomposition in computer-aided diagnosis.
• To improve the understanding of relationships between vessel geometries and arterial pathology.

Main Methods:

• Vascular tree decomposition using a weighted approximate convex decomposition (WACD) method.
• Incorporation of branching factor for structural decomposition and concavity analysis for aneurysm sites.
• Graph construction with edge weights derived from shape indices, followed by mesh decimation for clustering.

Main Results:

• WACD demonstrated effective decomposition, correctly clustering 84.7% of vessel sections compared to expert ground truth.
• The method showed superior performance over approximate convex decomposition (ACD).
• Validation on aneurysm-containing sections yielded low mean error rates: 7.8% for volumetric and 10.4% for surface area comparisons.

Conclusions:

• Decomposition of cerebral vasculature from CT angiograms into optimal convex regions supports computer-assisted diagnosis.
• The developed WACD method effectively handles complex vessel structures, including bifurcations and aneurysms.
• The WACD method shows promising results for enhancing the analysis of vascular pathologies.