Investigative Visualization

As the concepts in illustrative visualization became mature enough, they became widely used in various applications. The concept of investigative visualization slowly emerged as a form of visualization for addressing specific questions that required collective use of various graphics concepts described in the earlier section on illustrative visualization. This corresponds to fourth generation systems.

The application of many 3D volume visualization techniques remained limited to specific purposes where the needs could not be fulfilled otherwise. The use of 3D visualization techniques was not widespread because of several potential factors: It usually required a certain amount of specialized prior knowledge about the technique, the computation time and the cost of required high-performance systems were both high. Imaging advances in medicine helped systems produce better 3D images by slicing the sampling space into sets of image planes, while the rendering techniques were tools for putting them back together to produce a 3D image that did not add any new information. Thus, volume rendering was considered to be clinically useful only in cases where the 3D morphology could provide a visualization advantage.

But soon, new forms of imaging techniques were developed that contained functional information associated with the anatomical information in the images. Such functional

FIGURE 2 Visualization of beating heart: interactive dynamic volume rending of cardiac cine-MR images. The cine frame snapshots show the contraction cycle. (Images courtesy of M. Solaiyappan, Tim Poston, Pheng Ann Heng, Elias Zerhouni, Elliot McVeigh.)

imaging techniques needed volume rendering to visualize the functional information that could not be visualized otherwise. Volume rendering benefited various functional visualization applications that required investigative imaging. To illustrate investigative visualization several examples are now presented. The difference between investigative visualization and illustrative visualization is not in their fundamental concepts, but in their specific applications.

Stereoscopic 3D Visualization

One of the most important developments in volume visualization that demonstrated the full potential of volume rendering is stereoscopic 3D rendering. A stereo pair (left and right-eye views) of plain volume rendered images could be viewed using stereo goggles and provided the necessary left- and right-eye interleaving to produce the virtual 3D volume rendered image. Stereoscopic visualization represents a milestone in volume visualization in terms of promoting its popularity and acceptance. The traditional display methods that present a 3D object on a 2D display do not make use of the depth perception as a powerful visual cue to convey the 3D spatial information. Another powerful advantage of stereoscopic 3D visualization is its inherent ability to let the human eye play the role of a 3D filter to suppress the effect of noise in the image. Noise in images such as MRI volume usually appears around structures of interest. When the rendered image is presented in 2D, the noise appears projected on the structures and obscures them. But when the data are presented in 3D stereo, noise is resolved by depth perception and poses little distraction when the eye focuses on the structure of interest. The computational load doubles in stereoscopic 3D rendering because it involves creating a pair of views.

Dynamic Visualization

One of the simplest functional imaging examples is the cine-MR technique used to image the beating of the heart (Figs 2 and 3). Displaying the slices as cine-frames provides the motion cues that represent the functional aspect of the beating heart. But in such a 2D display, the through-image-plane motion would be suppressed, amounting to moderation of actual motion. Thus, volume rendering as a real-time display could provide in one sequence all the necessary information that can describe the motion of the beating heart with its three-dimensional dynamic properties [35,36].

Developmental Visualization and 3D Volume Morphing

The study of 3D shape changes in morphology is a challenging area of investigation, but in many cases there may not be a

FIGURE 2 Visualization of beating heart: interactive dynamic volume rending of cardiac cine-MR images. The cine frame snapshots show the contraction cycle. (Images courtesy of M. Solaiyappan, Tim Poston, Pheng Ann Heng, Elias Zerhouni, Elliot McVeigh.)

FIGURE 3 Visualization of left ventricle strain fields (color-coded) combined with tagged MR volume. See also Plate 113.

unique way of deriving this information without making certain assumptions. Such assumptions may cause loss of generality, but in an investigative study, posing those assumptions and visualizing the corresponding results may provide a better understanding of the functional information present in the image. One such specific developmental visualization is in the field of craniofacial modeling and visualization. For example, given two different stages in the craniofacial development in children with normal and abnormal growth, it would be instructive to visualize how the normal and abnormal

FIGURE 3 Visualization of left ventricle strain fields (color-coded) combined with tagged MR volume. See also Plate 113.

growth functions would exhibit themselves if they were expressed in the opposite group (Fig. 4). To describe the growth function, usually homologous landmarks that can be uniquely identified in craniofacial morphology are provided in the two stages that describe the growth. However, because there are few such homologous points (about 60), the visualization problem here would be to interpolate the growth function over the 3D space that represents the volume [48]. Using such a growth function, one can then model the morphing of the volume rendered image to produce the developmental visualization. Thus, by applying the growth function to a different subject, one could get a qualitative view of the effect of pseudo

FIGURE 4 Visualization of volumetric morphing. Study of sagittal synostosis. Growth form derived from sagittal synostosis patients is applied on normal subject (left) to visualize the simulated synosotosis (right). (Images courtesy of (Joan Richtsmeier, Shiaofen Fang, R Srinivasan, Raghu Raghavan; M. Solaiyappan, Diana Hauser.)

FIGURE 4 Visualization of volumetric morphing. Study of sagittal synostosis. Growth form derived from sagittal synostosis patients is applied on normal subject (left) to visualize the simulated synosotosis (right). (Images courtesy of (Joan Richtsmeier, Shiaofen Fang, R Srinivasan, Raghu Raghavan; M. Solaiyappan, Diana Hauser.)

craniofacial development on the subject. Thus, developmental visualization is essentially a 3D morphing technique that uses biologically appropriate control functions to describe the morphing. Also, it is important to notice that these morphing techniques attempt to morph (i.e., translate, rotate, scale) every voxel in the volume, unlike the more conventional morphing techniques used for 3D surfaces. Volumetric morphing is computationally intensive; however, it can avoid topological problems, such as self-intersection, that 3D surface morphing can easily encounter.

0 0

Post a comment