The Center for Interactive Ray-Tracing and Photo Realistic Visualization
The SCI Institute is honored to host a new Center of Excellence for Interactive Ray-Tracing and Photo Realistic Visualization dedicated to the development of tools for interactively visualizing large-scale datasets on the fly with advanced lighting and material models that help users understand the subtle detail in high-fidelity datasets. The Center of Excellence program is funded by the state of Utah to accelerate the commercialization of promising technologies developed at the state's Universities.
Almost every modern computer comes with a graphics processing unit (GPU) that implements an object-based graphics algorithm for fast 3-D graphics. The object-based algorithm in these chips was developed at the University of Utah in the 1970s. While these chips are extremely effective for video games and the visualization of moderately sized models, they cannot interactively display many of the large models that arise in computer-aided design, film animation, and scientific visualization. Researchers at the University of Utah have demonstrated that image-based ray tracing algorithms are more suited for such large-scale applications. A substantial code base has been developed in the form of two ray tracing programs. The new Center aims to improve and integrate these programs to make them appropriate for commercial use.
Scientific visualization as currently understood and practiced is still a relatively new discipline. As a result, we visualization researchers are not necessarily accustomed to undertaking the sorts of self-examinations that other scientists routinely undergo in relation to their work. Yet if we are to create a disciplinary culture focused on matters of real scientific importance and committed to real progress, it is essential that we ask ourselves hard questions on an ongoing basis. What are the most important research issues facing us? What underlying assumptions need to be challenged and perhaps abandoned? What practices need to be reviewed? In this article, I attempt to start a discussion of these issues by proposing a list of top research problems and issues in scientific visualization. [PDF version]
CT Segmentation and 3-D Reconstruction with Morphometric Analysis for Evaluation of Occipital-Cervical Instability in Children
Lindsey J. Healy
Down syndrome is a common chromosomal disorder that affects 0.15 percent of the total population. Individuals with Down syndrome are prone to spinal instability due to congenital abnormalities in the occipital-cervical (O-C1) joint near the base of the skull. To determine the possible abnormality causing this instability, an image-processing pipeline was created by combining several available software packages and custom made software. Patients with Down syndrome and spinal instability at the O-C1 joint were age-matched with controls. The subject data was assessed and a congenital abnormality was defined in the superior articular facets of C1 in the Down syndrome patients. The software developed helped to visualize the abnormality and could be used in a clinical setting to help aide in the diagnosis and screening for spinal instability in Down syndrome patients.
Associate Professor Ross Whitaker and Assistant Professor David Breen (Drexel University) have been developing surface modeling techniques based on level sets. These techniques include efficient algorithms for representing level-set surfaces and a new formulation for deforming one 3D object into another (also called 3D morphing). This combination of technologies provides an easy and efficient way to smoothly transform any 3D object to the shape of a target form. Recently, these methods were adopted by the special effects gurus at Frantic Films to create stunning visual effects for the movie Scooby-Doo 2.
Visualization of Vortex Flow Structures in Aerodynamics Applications
Dr. Xavier Tricoche
Computational Fluid Dynamics (CFD) has become an essential tool in various engineering fields. In aeronautics it is a key element in the design of modern aircrafts. The performances of today's computers combined with the increasing complexity of physical models yields numerical simulations that accurately reproduce the flow structures observed in practical experiments and permit to study their impact on flight stability. Yet, to fully exploit the huge amount of information contained in typical data sets engineers require powerful post-processing techniques that allow insight into the results of their large-scale computation.
Flow visualization aims at addressing this challenge by offering intuitive and effective depictions of interesting flow patterns. Unfortunately, many problems remain that limit the usefulness of existing methods in practical applications. Our recent work has focused on the design of new visualization techniques suitable for large-scale CFD simulations. Special emphasis was put on critical flight situations that lead to turbulent and vortical flows as well as complex and structurally involved phenomena like flow recirculation and vortex breakdown.
Visualization of Anatomic Covariance Tensor Fields
Gordon L. Kindlmann, David M. Weinstein, Agatha D. Lee, Arthur W. Toga, Paul M. Thompson
The computation, visualization, and interpretation of brain variability remains a significant challenge in computational neuroanatomy. Current deformable registration methods can generate, for each vertex of a polygonal mesh modeling the cortical surface, a distribution of displacement vectors between the individual models and their average, which can be summarized as a covariance tensor. While analysis of anatomical covariance tensor fields promises insight into the structural components of aging and disease, basic understanding of the tensor field structure is hampered by the lack of effective methods to create informative and interactive visualizations. We describe a novel application of superquadric tensor glyphs to anatomic covariance tensor fields, supplemented by colormaps of important tensor attributes. The resulting visualizations support a more detailed characterization of population variability of brain structure than possible with previous methods, while also suggesting directions for subsequent quantitative analysis.
Figure 1: (a) Ellipsoid tensor glyphs, (b) Space of superquadrics, (c) Superquadric tensor glyphs. Ellipsoidal glyphs (a) suffer from visual ambiguity. The range of superquadrics (b) used for tensor glyphs is highlighted with the gray triangle. Superquadric glyphs (c) differentiate shape and convey orientation more clearly than do ellipsoids.
We have proposed a novel point-based approach to view dependent isosurface extraction. We also introduce a fast visibility query system for the view dependent traversal, which exhibits moderate memory requirements. Using this technique, we achieved an interactive interrogation of the full visible woman dataset (1GB) at more then four frames per second on a desktop computer. The point-based approach is based on an extraction scheme that classifies different sections of the isosurface into four categories. The classification is based on the size of the geometry when projected onto the screen. In particular, we use points to represent small and sub-pixel triangles, as well as large sections of the isosurface whose projection has sub-pixel size. An important issue raised by point-based processing is how to assign a normal to an isolated point representing a large, but far, section of the isosurface. We propose to define such normals during a post processing of the extracted isosurface and provide the corresponding hardware implementation.
Figure 1: Left: A section of the visible female skeleton. Middle: A closeup view of the extracted points. Right: The final visibility mask. The color represent different levels of the mask hierarchy
Diffusion tensor MRI visualization is a growing field of research. The scanners are collecting better data all the time, and doctors and scientists are constantly discovering new applications for this data. However, unlike scalar and vector data, high-dimensional tensors are not always intuitive to visualize. When devising new strategies for DT-MRI visualization, it is important to understand both what exactly it is that is being measured and what insights the doctors and scientists are hoping to gain from the data.
The success of diffusion magnetic resonance imaging (MRI) is deeply rooted in the powerful concept that during their random, diffusion-driven displacements, molecules probe tissue structure at a microscopic scale well beyond the usual image resolution. As diffusion is truly a three dimensional process, molecular mobility in tissues may be anisotropic, as in brain white matter. With diffusion tensor imaging (DTI), diffusion anisotropy effects can be fully extracted, characterized, and exploited, providing even more exquisite details on tissue microstructure. The most advanced application is certainly that of fiber tracking in the brain, which, in combination with functional MRI, might open a window on the important issue of connectivity. DTI has also been used to demonstrate subtle abnormalities in a variety of diseases (including stroke, multiple sclerosis, dyslexia, and schizophrenia) and is currently becoming part of many routine clinical protocols.
Gordon Kindlmann, Richard A. Normann, Arun Badi, Charles Keller, Greg M. Jones, Christopher R. Johnson
Biomedical applications of small animal imaging are creating exciting opportunities to extend the scientific impact of visualization research. Specifically, the effective pairing of non-linear image filtering and direct volume rendering is one strategy for scientists to quickly explore and understand the volumetric scans of their specimens. Microscopic computed tomography imaging is an increasingly popular and powerful modality for small animal imaging. Here we highlight early work from collaborations at the University of Utah between the Scientific Computing and Imaging (SCI) Institute and the Department of Bioengineering, and between the SCI Institute and the Division of Pediatric Hematology-Oncology in the Department of Pediatrics. In the first instance, volume rendering provides information about the three-dimensional configuration of an electrode array implanted into the auditory nerve of a feline. In the second instance, volume rendering shows promise as a tool for visualizing bone tissue in the mouse embryo, although the signal-to-noise characteristics of the data require the use of sophisticated image pre-processing.
The data for both of these investigations was acquired with a General Electric EVS RS-9 computed tomography scanner at the University of Utah Small Animal Imaging Facility. The scanner generates 16-bit volumes roughly one gigabyte in size, with a spatial resolution of 21 x 21 x 21 microns.
Jennifer Simpson, Eric luke, MS., Allen Sanderson, Phd.
In the last few years, scientists and researchers have given a great deal of attention to the area of remote visualization of scientific datasets within collaborative environments. This recent interest has been fueled by the use of interactive viewing as the primary means by which researchers explore large datasets. However, researchers often need to extend this interactivity in order to collaborate with colleagues who are geographically separated. Most current remote visualization tools allow multiple parties to view images from different locations, but pose problems with efficiency and user interactivity.
Undergraduate research opportunities at the University of Utah have dramatically increased in the past several years. Positions that were once reserved for graduate students have now been opened up through a variety of programs. The SCI Institute has taken full interest in promoting and co-sponsoring these opportunities--namely, the Engineering Scholars Program and the Access Program for Women in Science and Mathematics, both of which offer scholarships to first year undergraduates and present them with the opportunity to work side-by-side with professors and researchers in various labs.
Star-Ray Interactive Ray-Tracer Debuts at SIGGRAPH 2002
Michelangelo's David; a large-scale seismic volume; the moons of Jupiter; four half-full wine glasses; the Greek Parthenon. What do these models have in common? They were all part of the Star-Ray demonstration at SIGGRAPH 2002, held in San Antonio, July 22-24. With support from SGI, researchers from SCI, GDC, VisSim, and the Graphics group designed, implemented, and premiered interactive ray-tracing of a complex virtual underwater world for thousands of graphics enthusiasts at the world's premier computer graphics conference.
(left to right) Living Room Scene, Graphics Museum, Science Room, Galaxy Room, Atlantis Scene
Parametric Method for Correction of Intensity Inhomogeneity in MRI Data
A. Samsonov, R. Whitaker, C.R. Johnson
Intensity inhomogeneity is one of the main obstacles for MRI data post processing. The problem requires retrospective correction due to the strong dependence of the inhomogeneity on patient anatomy and the accompanying acquisition protocol. We have developed a new method for correcting the inhomogeneities using a polynomial estimation of the bias field. The method minimizes the composite energy function to find parameters of the polynomial model. The energy function is designed to provide a robust estimation of the bias field by combining measures from histogram analysis and local gradient estimation. The method was validated on a wide range of MRI data obtained with coils of different types and under different acquisition protocols.
The developed method provides reliable estimation of the intensity inhomogeneities in MRI data. The correction times are dependant on the number of parameters the model used, the dataset size and the degree of subsampling in estimation of both local and global terms and vary from 1 to 5 minutes using a mid range PC.
Traditionally, processing tomographic data begins with reconstructing volumes. However, when the tomographic data is incomplete, noisy, or misregistered tomographic reconstruction can produce artifacts in the volume, which makes subsequent segmentation and visualization more difficult. Researchers in the SCI institute are developing direct methods for segmenting tomographic data. The strategy is to fit 3D surface models directly to the tomographic projects, rather than the volume reconstructions. In this way, the surface fitting is not influenced by reconstruction artifacts. Implementing this strategy requires several technical advances. First is a mathematical formulation that relates object shape directly to tomographic projections. This results in a description of how surfaces should deform in order to match the tomographic data. The second advance is the use of a surface modeling technology that can accommodate a wide variety of shapes and support incremental deformations. This is done using 3D level-set models, which results in a 3D partial differential equation (PDE). The final advance is development of computational schemes that allow us to solve these PDE's efficiently. For these we have developed the incremental projection method which significantly reduces the amount of computation needed to deform these 3D surface models.
The goal of visualization is to aid in the understanding of complex scientific data, typically using techniques from the fields of computer graphics and animation. To gain additional insight, immersive visualization places the user directly within the data space through virtual reality technology. The resulting immersive experience allows exploration from a first-person perspective, in contrast to the third-person interaction typical of desktop environments.
The feeling of actually "being there" in a virtual environment, also known as presence, is created by fooling one or more of the user's senses with computer generated cues. In a typical system, stereo images provide a sense of visual depth and natural interaction is achieved with tracking sensors and 3D input devices. More advanced systems may also include force-feedback, spatialized audio, and/or voice recognition capabilities to increase the sense of presence.
Imagine directly navigating through a scientific dataset, much like one experiences the real world. How would it feel to investigate the interesting features of the data with your sense of touch? Would this capability be useful or not? These are the types of questions Virtual Reality (VR) researchers at the SCI institute currently seek to answer.
This page documents the evolution of our volume rendering project, named Simian. As volume rendering goes, this system is quite a departure from the way this process is typically approached. Most direct volume renderings produced today employ one-dimensional transfer functions, which assign color and opacity to the volume based solely on the single scalar quantity that comprises the dataset. Multi-dimensional transfer functions, however, are an effective way to extract specific material boundaries and convey subtle surface properties. However, finding good transfer functions is hard enough in one dimension, let alone two or three.
SCIRun/BioPSE: Problem Solving Environments for the Next Generation of Scientific Computing
In March of 2001, the Scientific Computing and Imaging Institute reached a milestone with the long anticipated release of both SCIRun 1.0.0
and BioPSE 1.0.0.
"I am excited about the SCIRun and BioPSE software releases," says the Institute's director Christopher R. Johnson. "We have been working on SCIRun since 1992. What started out as software designed by a few people (Steve Parker with help from David Weinstein) has grown into a substantial software effort with more than 50 contributors." Johnson continues, "BioPSE has been a focused software effort since 1999. I am thrilled to see SCIRun and BioPSE released and look forward to seeing how scientists and engineers use these software packages to solve their application domain problems."
If you have played just about any modern Nintendo(tm) or Playstation (tm) computer game, then you have encountered meshes. Many games make heavy use of what are called polygonal surface meshes, or surfaces built up out of polygons. They are used in the models of many of the people and cars and other 3D things within the game. Polygonal models are a good way for the game designers to get what they want out of the hardware inside the computer or whatever you are using to play the game. I'll explain what a mesh actually is, how it is constructed, and how engineers use meshes to solve problems.
By Oleg Portniaguine. An interesting, but very challenging kind of imaging is to visualize the interior of a non-transparent object (such as a human body) using physical fields measured outside the body. This imaging is achieved through a mathematical engine known as "inverse problems solving" or "statistical optimization", one of the key research directions at the SCI Institute.
Another research direction being pursued at the SCI Institute is solving "ill-posed" imaging problems by constraining the solution with focusing criterion. This allows us to reconstruct sharp solutions out of smooth data. The process is made possible by selecting special stabilizing functions that permit sharp solutions. Applications of this technique range from bioelectric source localization utilizing magneto and electro -encephalography data for medical imaging, to geophysical inversions with gravity and magnetic fields.
The technology of Magnetic Resonance Imaging (MRI) has been used in an ever-increasing variety of applications in the area of medical imaging. This is partly because of MRI's basic ability to non-invasively and non-destructively take images of living tissue, and also because of the inherent flexibility in the way that MRI machines are programmed in order to acquire images. One relatively new method of using MRI technology is called "diffusion tensor imaging". By measuring the directions along which water molecules diffuse through brain tissue, this technology allows us to explore the structure of the brain in new ways which benefit many disciplines. For example, it can help doctors to better detect abnormalities in brain tissue, cognitive scientists to better understand the interconnections between the functional units of the brain, and show biologists how brain tissue becomes organized in a growing fetus.