Level-set-based artery-vein separation in blood pool agent CE-MR angiograms

Area of Science:

• Medical Imaging
• Image Analysis
• Cardiovascular Imaging

Background:

• Blood pool agents (BPAs) in contrast-enhanced magnetic-resonance angiography (CE-MRA) enable prolonged imaging for enhanced contrast and resolution.
• Steady-state imaging with BPAs allows complete vascular system distribution but causes simultaneous arterial and venous enhancement, hindering interpretation.

Purpose of the Study:

• To present a semiautomated method for artery-vein separation from steady-state BPA CE-MRA data.
• To improve the visualization of arteries and veins in CE-MRA by separating them effectively.

Main Methods:

• A level-set framework is employed, initializing arterial and venous surfaces from central axes.
• Simultaneous surface evolution captures arterial and venous vasculature, using arrival times to resolve proximity-based leakage.
• External forces (intensity, gradient, vessel-enhancement) and internal smoothness constraints guide the level-set evolution.

Main Results:

• Investigated the robustness and accuracy of different external forces and their combinations on seven patient datasets.
• Compared level-set segmentation results against manually obtained reference segmentations.
• Identified a combination of intensity- and gradient-based forces with a curvature-based smoothness constraint as yielding the best results.

Conclusions:

• The developed method achieves accurate artery-vein separation in BPA CE-MRA with minimal user interaction.
• This technique significantly enhances the visualization of both arterial and venous vasculature.
• The findings support improved diagnostic capabilities in cardiovascular imaging using CE-MRA.