Christopher R. Johnson

Top scientific visualization research problems

Scientific visualization as currently understood and practiced is still a relatively new discipline. As a result, we visualization researches 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 creat 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 needs 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.

A Next Step: Visualizing Errors and Uncertainty

The development of formal theoretical frameworks and the creation of new visual representations of error and uncertainty will be fundamental to a better understanding of 3D experimental and simulation data. Such improved understanding will validate new theoretical models, enable better understanding of data, and facilitate better decision making. We urge the scientific visualization research community to take the next step and make visually representing errors and uncertainties the norm rather than the exception.

Isosurfacing in span space with utmost efficiency (ISSUE)

We present efficient sequential and parallel algorithms for isosurface extraction. Based on the Span Space data representation, new data subdivision and searching methods are described. We also present a parallel implementation with an emphasis on load balancing. The performance of our sequential algorithm to locate the cell elements intersected by isosurfaces is faster than the Kd tree searching method originally used for the Span Space algorithm. The parallel algorithm can achieve high load balancing for massively parallel machines with distributed memory architectures.

The SCIRun computational steering software system

We present the design, implementation and application of SCIRun, a scientific programming environment that allows the interactive construction, debugging, and steering of large-scale scientific computations. Using this “computational workbench,” a scientist can design and modify simulations interactively via a dataflow programming model. SCIRun enables scientists to design and modify model geometry, interactively change simulation parameters and boundary conditions, and interactively visualize geometric models and simulation results. We discuss the ubiquitous roles SCIRun plays as a computational tool (e.g. resource manager, thread scheduler, development environment), and how we have applied an object oriented design (implemented in C++) to the scientific computing process. Finally, we demonstrate the application of SCIRun to large scale problems in computational medicine.

Virtual Histology of Transgenic Mouse Embryos for High-Throughput Phenotyping

A bold new effort to disrupt every gene in the mouse genome necessitates systematic, interdisciplinary approaches to analyzing patterning defects in the mouse embryo. We present a novel, rapid, and inexpensive method for obtaining high-resolution virtual histology for phenotypic assessment of mouse embryos. Using osmium tetroxide to differentially stain tissues followed by volumetric X-ray computed tomography to image whole embryos, isometric resolutions of 27 μm or 8 μm were achieved with scan times of 2 h or 12 h, respectively, using mid-gestation E9.5–E12.5 embryos. The datasets generated by this method are immediately amenable to state-of-the-art computational methods of organ patterning analysis. This technique to assess embryo anatomy represents a significant improvement in resolution, time, and expense for the quantitative, three-dimensional analysis of developmental patterning defects attributed to genetically engineered mutations and chemically induced embryotoxicity.

Uintah: a massively parallel problem solving environment

Describes Uintah, a component-based visual problem-solving environment (PSE) that is designed to specifically address the unique problems of massively parallel computation on tera-scale computing platforms. Uintah supports the entire life-cycle of scientific applications by allowing scientific programmers to quickly and easily develop new techniques, debug new implementations and apply known algorithms to solve novel problems. Uintah is built on three principles: (1) as much as possible, the complexities of parallel execution should be handled for the scientist, (2) the software should be reusable at the component level, and (3) scientists should be able to dynamically steer and visualize their simulation results as the simulation executes. To provide this functionality, Uintah builds upon the best features of the SCIRun (Scientific Computing and Imaging Run-time) PSE and the DoE (Department of Energy) Common Component Architecture (CCA).

Ensemble-Vis: A Framework for the Statistical Visualization of Ensemble Data

Scientists increasingly use ensemble data sets to explore relationships present in dynamic systems. Ensemble data sets combine spatio-temporal simulation results generated using multiple numerical models, sampled input conditions and perturbed parameters. While ensemble data sets are a powerful tool for mitigating uncertainty, they pose significant visualization and analysis challenges due to their complexity. In this article, we present Ensemble-Vis, a framework consisting of a collection of overview and statistical displays linked through a high level of interactivity. Ensemble-Vis allows scientists to gain key scientific insight into the distribution of simulation results as well as the uncertainty associated with the scientific data. In contrast to methods that present large amounts of diverse information in a single display, we argue that combining multiple linked displays yields a clearer presentation of the data and facilitates a greater level of visual data analysis. We demonstrate our framework using driving problems from climate modeling and meteorology and discuss generalizations to other fields.

Independent component analysis for EEG source localization

We consider a spatiotemporal method for source localization, taking advantage of the entire EEG time series to reduce the configuration space we must evaluate. The EEG data are first decomposed into signal and noise subspaces using a principal component analysis (PCA) decomposition. This partitioning allows us to easily discard the noise subspace, which has two primary benefits: the remaining signal is less noisy, and it has lower dimensionality. After PCA, we apply independent component analysis (ICA) on the signal subspace. The ICA algorithm separates multichannel data into activation maps due to temporally independent stationary sources. For each activation map we perform an EEG source localization procedure, looking only for a single dipole per map. By localizing multiple dipoles independently, we substantially reduce our search complexity and increase the likelihood of efficiently converging on the correct solution.

The effects of inhomogeneities and anisotropies on electrocardiographic fields: a 3-D finite-element study

The aim of this study was to quantify the effects of selected inhomogeneities and anisotropies on computed electric potential fields associated with the electrocardiographic forward problem. The model construction was based on the Utah Torso model and included geometry for major anatomical structures such as subcutaneous fat, skeletal muscle, and lungs, as well as for epicardial fatpads, major arteries and veins, and the sternum, ribs, spine, and clavicles. Measured epicardial potentials served as the electrical source for solutions to the electrocardiographic forward problem computed using the finite element method (FEM). The geometry of the torso model for each simulation was constant, but different combinations of conductivities were assigned to individual organs or tissues. Comparisons of different conductivity combinations followed one of two basic schemes: 1) a homogeneous torso served as the reference against which we compared simulations with a single organ or tissue and assigned its nominal conductivity, or 2) a fully inhomogeneous torso served as the reference and we removed the effect of individual organs or tissues by assigning it the homogeneous conductivity value. When single inhomogeneities were added to an otherwise homogeneous isotropic model, anisotropic skeletal muscle (at a 15:1 anisotropy ratio) and the right and left lung had larger average effects (12.8, 12.7, and 12.1% relative error (RE), respectively) than the other inhomogeneities tested. Our results for removing single inhomogeneities show that the subcutaneous fat, the anisotropic skeletal muscle (with the degree of anisotropy equal to 7:1), and the lungs have larger average impacts on the body surface potential distributions than other elements of the model (with values of 14.9, 12.6, and 11.7% RE, respectively). The results also show that the size of the effect depended strongly on the distribution of epicardial potentials. The results of this study suggest that accurate representation of tissue inhomogeneity has a significant effect on the accuracy of the forward solution, with regions near the torso surface playing a larger role, in general, than those near the heart.

Noise-adaptive nonlinear diffusion filtering of MR images with spatially varying noise levels.

Anisotropic diffusion filtering is widely used for MR image enhancement. However, the anisotropic filter is nonoptimal for MR images with spatially varying noise levels, such as images reconstructed from sensitivity‐encoded data and intensity inhomogeneity‐corrected images. In this work, a new method for filtering MR images with spatially varying noise levels is presented. In the new method, a priori information regarding the image noise level spatial distribution is utilized for the local adjustment of the anisotropic diffusion filter. Our new method was validated and compared with the standard filter on simulated and real MRI data. The noise‐adaptive method was demonstrated to outperform the standard anisotropic diffusion filter in both image error reduction and image signal‐to‐noise ratio (SNR) improvement. The method was also applied to inhomogeneity‐corrected and sensitivity encoding (SENSE) images. The new filter was shown to improve segmentation of MR brain images with spatially varying noise levels. Magn Reson Med 52:798–806, 2004.