Next: Results Up: Image Analysis Methods Previous: Quantitative Measures of

3D Visualization

Here, we briefly discuss our approach to the complex process of visualizing the 4D results. This includes a need to visualize intermediate information that is found within our algorithms, as well as the final motion parameters described above. Please see next section for examples.

The shape-based algorithm is closely related to the differential characteristics of the LV surfaces throughout the cardiac cycle, and the interpretation of this information is difficult when only looking at the numerical data produced by the curvature estimation process. Thus the visual inspection of the spatial distribution of the curvatures or their combinations can be of great help to choose proper shape matching properties or to assess their dynamic behavior over time. To fulfill this need, we use the SUNVISION package along with our custom-written shading program, which assigns a color to each point on the surface according to its curvature value, and finds the color of any point inside each triangular facet by interpolating the colors of the three vertices. The shading operation is based on Phong's illumination model, and it conveys the depth information and allows to render the morphological details of the surfaces.

To display the single surface motion parameters, we use arrows emanating from the surface of time and ending at the surface of time to represent the flow vectors.

To visualize the cross-wall measures such as thickness vectors and strain cubes, we need to display the endocardium and epicardium as well as the cross-wall measures simultaneously. Each of these three entities is assigned a different texture: the endocardial surface has an opaque red appearance, the epicardial surface is transparent green, while the cross-wall vectors or cubes appear opaque blue.

While static visualization presents us with information of the curvature properties and motion parameters, dynamic display is a far more efficient approach for the viewing of shape-based motion tracking. Hence, movies of the dynamic curvature maps, motion vectors, cross-wall thickness changes, and cube deformation have been assembled, and will be presented.

Next: Results Up: Image Analysis Methods Previous: Quantitative Measures of

Wed Feb 23 15:02:52 EST 1994