Assessment of left ventricle dynamic from cardiac magnetic resonance imaging by means a correspondence approach

Abstract

Abstract In this research, an approach to assess the heart dynamics is reported. The cardiac magnetic resonance images are considered for assessing the left ventricle motion and deformation. The shape of the cavity is obtained by means a segmentation procedure based on a clustering algorithm at an initial instant. This three-dimensional structure is used to establish a region of interest around the border of the structure. An optical flow method allows determining the displacement vector of this region and then defining the shapes of the cavity during the all cardiac cycle. The points of the left ventricle are followed using the displacement vectors in the cardiac cycle, obtaining thus, a dense motion field of the cavity. This approximate deformation field is refined with a correspondence method that working in the three-dimensional space. The mapping of the points that define the left ventricle in the cardiac cycle obtained with the correspondence procedure are then used for computing a set of clinical parameters that allows assessing the motion and deformation of this principal structure of the human heart. The torsion, radial and longitudinal contraction are quantified. The obtained results are promising for evaluating the heart dynamics.