"Visualization" refers to rendering techniques
that make images of non-geometric data sets.
The data may be a set of "voxels,"
which contain a density for each element of a cubic subdivision of a volume.
Alternatively, the data may be a series of images
from a medical recording device.
Sometimes, data from several such devices can be combined
to provide a more complete image of the phenomemon that is being studied.
These images show some of the past work in visualization at the PCG.
Electron Microscopy Visualization
As part of the
Collaboratory for Microscopic Digital Anatomy,
we contributed to significant improvements in
electron microscope (EM) tomography visualization.
EM is an important bioligical/medical research tool,
but state-of-the-art facilites are expensive and scarce.
This image of a dendrite was generated using a tomographic reconstruction
from a series of EM images.
Gordon Kindlmann and James W. Durkin,
Semi-Automatic Generation of Transfer Functions for Direct Volume Rendering,
IEEE Symposium on Volume Rendering Proceedings, October, 1998, pp. 79-86 & 170
Stephen J. Young, Gary (Guo) You Fan, David Hessler, Steve Lamont, Todd T. Elvins,
Martin Hadida-Hassan, Gary Hanyzewski, James W. Durkin, Philip M. Hubbard,
Gordon L. Kindlmann, Eric Chih-Cheng Wong, Donald P. Greenberg, Sidney Karin,
and Mark Ellisman,
Implementing a Collaboratory For Microscopic Digital Anatomy,
Int. J. Supercomputing Applications and High Performance Computing, (10)2-3, Sum-Fall, 1996, pp. 170 - 181
Intravascular Ultrasound Imaging
Intravascular ultrasonography and x-ray angiography
provide two complimentary techniques for imaging the moving coronary arteries.
While a student at the PCG, Jed Lengyel developed techniques that combine the strengths of both.
The moving three-dimensional arterial tree is recovered from
a stereo pair of angiograms through the use of compound-energy "snakes."
The intravascular ultrasound slices are then placed
at their proper positions in time and space,
and the combined data is dynamically displayed.
Past techniques have assumed that the ultrasound slices are parallel
and that the vessel being imaged is static and straight.
We have been able to incorporate the curved arterial tree,
and to display the data with reference to the dynamic angiogram projections.
Jed Lengyel, Donald P. Greenberg, and Richard Popp,
Time-Dependent Three Dimensional Intravascular Ultrasound,
Computer Graphics (SIGGRAPH '95 Conference Proceedings), Vol. 29, August, 1995, pp. 457-464
Volume Reconstruction Filters
This collage of four images was generated by
Steve Marschner and Richard Lobb for the 1994 paper
An Evaluation of Reconstruction filters for Volume Rendering.
The image depicts an analytic function in comparison to
three methods of reconstructing the function from a sampling set.
One can see that the traditional trilinear interpolation method
fails miserably, and that
the cosine windowed sinc function improves upon
the results from a cubic B-spline reconstruction.
Stephen R. Marschner and Richard J. Lobb,
An Evaluation of Reconstruction Filters for Volume Rendering,
Proceedings of IEEE Visualization Conference, 1994, pp. 100-109
|
|
|