Quantification of cross-sectional artery wall motion with IVUS image registration
Atherosclerotic lesions have been shown to have different mechanical properties than the non-diseased artery. Calculating vessel wall strain from cross-sectional vessel wall motions allows for the measurement of local stiffness. In this paper, a robust method is developed to track cross-sectional displacements of an artery wall using two intravascular ultrasound (IVUS) images acquired at two different pressure levels respectively. First, the vessel wall region in each image is segmented semi-automatically by refining two spline-based contours to the locations of inner and outer vessel wall borders. Then the ringlike wall region in one image is registered to its counterpart in the other image in polar coordinates. The registration is performed by minimizing an energy function of the 2D motion field based on a spline-based deformable model. Both intensity and gradient information of the images are used to construct the energy function so that an accurate registration can be achieved. Registration accuracy was tested on simulated motions using IVUS images of a human coronary artery and a porcine carotid. The wall displacement fields calculated from real motion images are also demonstrated