Using the methodology described above, the initial experimentation with the 4D shape-based motion tracking approach is reported here. To date, we have acquired six sets of 4D MRI data under the acute infarct animal model. In addition, we also have obtained data from five studies acquired from the 4D real-time Dynamic Spatial Reconstructor(DSR) at the Mayo Clinic[31]. The results here are from one of the MR studies. Experiments on other MR studies and DSR data have been underway.
Fig.1 shows a Delaunay tessellated endocardial surface.
Fig.2 shows the computed Gaussian, mean, and two principal curvature of an endocardial surface, while Fig.3 shows the bending energy() of the same surface. Fig.4 demonstrates the effects of the curvature-smoothing procedure. Fig.5 and Fig.6 display the bending energy map's temporal progress over four frames for normal and infarcted heart, respectively. Note the normal heart has more shape changes, hence more local deformation, than the infarcted one.
Fig.7 shows a dense set of motion vectors emanating from part of the endocardial surface. Fig.8 and Fig.9 present the motion trajectories of several endocardial points from ED to ES for a normal and an infarcted heart, respectively. The green trajectories represent the path of the registered points pre- and post- occlusion. Note that normal heart has larger trajectory path length(9.6mm) and ED-ES distance(6.7mm) than the infarcted one(7.9mm and 2.3mm, respectively).
Fig.10 is a 3D display of the LV wall: the opaque red endocardium and the transparent green epicardium, and also several opaque blue cross-wall cubes used for local strain measure. Fig.11 and Fig.12 show the deformation of one cube from ED to ES for normal and infarcted heart, respectively. There is again a clear difference between the two.