Vortex segmentation from cardiac MR 2D velocity images using region growing about vortex centres

Abstract

We present a new method for extracting vortex structures from 2D blood flow field images. The new technique first identifies vortex centres in the velocity field, and then grows regions about these points using a criterion based on the Jacobeans evaluated at the candidate points, and the size and shape of the region already determined. A convexity constraint is also employed to enforce coherency in the resulting regions. The algorithm accounts for the asymmetry and nonlinearity in vortex structures while maintaining the concept of a vortex centre. We present results of the method obtained from synthetic data and a blood flow velocity image.