Abstract #3825

Automatic Segmentation of 3D Phase Contrast MRI Using Velocity Guided Gradient Vector Flow

Robert L. Janiczek¹, Frederick H. Epstein^1,2, Scott T. Acton^1,3

¹Biomedical Engineering, University of Virginia, Charlottesville, VA, USA; ²Radiology, University of Virginia, Charlottesville, VA, USA; ³Electrical Engineering, University of Virginia, Charlottesville, VA, USA

Hemodynamic measurements using phase contrast MRI in mouse models of atherosclerosis provide insight into the molecular mechanisms of the disease. Automatic segmentation using active models could eliminate the need for manual segmentation prior to calculation of hemodynamic parameters such as wall shear stress. Traditionally, active model external forces have relied exclusively on the image magnitude. We propose a new active model external force that incorporates velocity data into the gradient vector flow (GVF) framework. Velocity guided GVF acts to push an active model along the direction of flow and is shown to improve segmentation results when compared to GVF.