Search

We present GENeVis, an application to visualize gene expression time series data in a gene regulatory network context. This is a network of regulator proteins that regulate the expression of their respective target genes. The networks are represented as graphs, in which the nodes represent genes, and the edges represent interactions between a gene and its targets. GENeVis adds features that are currently lacking in existing tools, such as mapping of expression value and corresponding p-value (or other statistic) to a single visual attribute, multiple time point visualization, and visual comparison of multiple time series in one view. Various interaction mechanisms, such as panning, zooming, regulator and target highlighting, data selection, and tooltips support data analysis and exploration. Subnetworks can be studied in detail in a separate view that shows the network context, expression data plots, and tables containing the raw expression data. We present a case study, in which gene expression time series data acquired in-house are analyzed by a biological expert using GENeVis. The case study shows that the application fills the gap between present biological interpretation of time series experiments, performed on a gene-by-gene basis, and analysis of global classes of genes whose expression is regulated by regulator proteins. [ABSTRACT FROM AUTHOR]

To make possible a more rigorous understanding of animal gene regulatory networks, the Berkeley Drosophila Transcription Network Project (BDTNP) has developed a suite of methods that support quantitative, computational analysis of three-dimensional (3D) gene expression patterns with cellular resolution in early Drosophila embryos. Defining the pattern of gene expression is an essential step toward further analysis in order to derive knowledge about the characteristics of gene expression patterns and to identify and model gene inter-relationships. To address this challenging task we have developed an integrated, interactive approach toward pattern segmentation. Here, we introduce a ridge-detection-based 3D gene expression pattern segmentation algorithm. We compare this algorithm to common 2D pattern segmentation methods, such as thresholding and edged-detection-based methods, which we have adapted to 3D pattern segmentation. We show that such automatic strategies can be improved to obtain better segmentation results by user interaction and additional post-processing steps. [ABSTRACT FROM AUTHOR]

The increasing complexity of models for prediction of the native spatial structure of RNA molecules requires visualization methods that help to analyze and understand the models and their predictions. This paper improves the visualization method for sequences of barrier trees previously published by the authors. The barrier trees of these sequences are rough topological simplifications of changing folding landscapes — energy landscapes in which kinetic folding takes place. The folding landscapes themselves are generated for RNA molecules where the number of nucleotides increases. Successive landscapes are thus correlated and so are the corresponding barrier trees. The landscape sequence is visualized by an animation of a barrier tree that changes with time. The animation is created by an adaption of the foresight layout with tolerance algorithm for dynamic graph layout problems. Since it is very general, the main ideas for the adaption are presented: construction and layout of a supergraph, and how to build the final animation from its layout. Our previous suggestions for heuristics lead to visually unpleasing results for some datasets and, generally, suffered from a poor usage of available screen space. We will present some new heuristics that improve the readability of the final animation. [ABSTRACT FROM AUTHOR]

Surface models of biomolecules have become crucially important for the study and understanding of interaction between biomolecules and their environment. We argue for the need for a detailed understanding of biomolecular surfaces by describing several applications in computational and structural biology. We review methods used to model, represent, characterize, and visualize biomolecular surfaces focusing on the role that geometry and topology play in identifying features on the surface. These methods enable the development of efficient computational and visualization tools for studying the function of biomolecules. [ABSTRACT FROM AUTHOR]

Based on the results of an interdisciplinary research project the paper addresses the embedding of knowledge about the function of different parts/structures of a macro molecule (protein, DNA, RNA) directly into the 3D model of this molecule. Thereby the 3D visualization becomes an important user interface component when accessing domain-specific knowledge — similar to a web browser enabling its users to access various kinds of information. In the prototype implementation — named BioBrowser — various information related to bio-research is managed by a database using a fine-grain access control. This also supports restricting the access to parts of the material based on the user privileges. The database is supplied by a SOAP web service so that it is possible (after identifying yourself by a login procedure of course) to query, to change, or to add some information remotely by just using the 3D model of the molecule. All these actions are performed on sub structures of the molecules. These can be selected either by an easy query language or by just picking them in th 3D model with the mouse. [ABSTRACT FROM AUTHOR]

Orbital fat, or the fat behind the eye, plays an important role in eye movements. In order to gain a better understanding of orbital fat mobility during eye motion, MRI datasets of the eyes of two healthy subjects were acquired respectively in seven and fourteen different directions of gaze. After semi-automatic rigid registration, the Demons deformable registration algorithm was used to derive time-dependent three-dimensional deformation vector fields from these datasets. Visualisation techniques were applied to these datasets in order to investigate fat mobility in specific regions of interest in the first subject. A qualitative analysis of the first subject showed that in two of the three regions of interest, fat moved half as much as the embedded structures. In other words, when the muscles and the optic nerve that are embedded in the fat move, the fat partly moves along with these structures and partly flows around them. In the second subject, a quantitative analysis was performed which showed a relation between the distance behind the sciera and the extent to which fat moves along with the optic nerve. [ABSTRACT FROM AUTHOR]

The bioelectric activity that takes place throughout the human body enables the action of muscles and the transmission of information in nerves. A variety of imaging modalities have therefore been developed to assess this activity. In particular, electrocardiography for the heart and electrocardiography and magnetoencephalography for the brain permit to measure non-invasively the resulting electric signal. Beyond their obvious clinical applications these techniques also open the door to a computational reconstruction of the physiological activity at the origin of this signal through the numerical solution of so-called inverse problems. In this case as well as in basic bioengineering research effective postprocessing tools are necessary to facilitate the interpretation of measured and simulated bioelectric data and permit to derive anatomical and functional insight from it. In contrast to scalar quantities for which effective depictions generally exist and are routinely used, the vector-valued nature of this information bears specific challenges that are insufficiently addressed by the visualization tools typically available to biomedical practitioners. This paper reports on the application of advanced vector visualization techniques to the postprocessing analysis of bioelectric fields as they arise in cardiovascular and inverse source reconstruction research. Our work demonstrates the ability of the corresponding visual representations to improve the interpretation of the data and support new insight into the underlying physiology. [ABSTRACT FROM AUTHOR]

For visualization of medical diffusion data one typically computes a tensor field from a set of diffusion volume images scanned with different gradient directions. The resulting diffusion tensor field is visualized using glyph- or tracking-based approaches. The derivation of the tensor, in general, involves a loss in information, as the n > 6 diffusion values for the n gradient directions are reduced to six diverse entries of the symmetric 3 × 3 tensor matrix. We propose a direct diffusion visualization approach that does not operate on the diffusion tensor. Instead, we assemble the gradient vectors on a unit sphere and deform the sphere by the measured diffusion values in the respective gradient directions. We compute a continuous deformation model from the few discrete directions by applying several processing steps. First, we compute a triangulation of the spherical domain using a convex hull algorithm. The triangulation leads to neighborhood information for the position vectors of the discrete directions. Using a parameterization over the sphere we perform a Powell-Sabin interpolation, where the surface gradients are computed using least-squares fitting. The resulting triangular mesh is subdivided using a few Loop subdivision steps. The rendering of this subdivided triangular mesh directly leads to a glyph-based visualization of the directional diffusion measured in the respective voxel. In a natural and intuitive fashion, our deformed sphere visualization can exhibit additional, possibly valuable information in comparison to the classical tensor glyph visualization. [ABSTRACT FROM AUTHOR]

We describe our recent work in applying physically-based global illumination to fiber tractography. The geometry of the fiber tracts is derived from diffusion tensor magnetic resonance imaging (DT-MRI) datasets acquired from the white matter in the human brain. Most visualization systems display such fiber tracts using local illumination, a rendering technology provided by the video card on a typical desktop computer. There is indirect evidence that the human visual system perceives the shape of a fiber more quickly and more accurately when physically-based illumination is employed. [ABSTRACT FROM AUTHOR]

Segmentation and surface extraction from 3D imaging data is an important task in medical applications. When dealing with scalar data such as CT or MRI scans, a simple thresholding in form of isosurface extraction is an often a good choice. Isosurface extraction is a standard tool for visualizing scalar volume data. Its generalization to color data such as cryosections, however, is not straightforward. In particular, the user interaction in form of selection of the isovalue needs to be replaced by the selection of a three-dimensional region in feature space. We present a user-friendly tool for segmentation and surface extraction from color volume data. Our approach consists of several automated steps and an intuitive mechanism for user-guided feature selection. Instead of overburden the user with complicated operations in feature space, we perform an automated clustering of the occurring colors and suggest segmentations to the users. The suggestions are presented in a color table, from which the user can select the desired cluster. Simple and intuitive refinement methods are provided, in case the automated clustering algorithms did not immediately generate the desired solution exactly. Finally, a marching technique is presented to extract the boundary surface of the desired cluster in object space. [ABSTRACT FROM AUTHOR]

Arising from the clinical need for multimodal imaging, an integrated system for automated multimodal image registration and multi-source volume rendering has been developed, enabling simultaneous processing and rendering of image data from structural and functional medical imaging sources. The algorithms satisfy real-time data processing constraints, as required for clinal deployment. The system represents an integrated pipeline for multimodal diagnostics comprising of multiple-source image acquisition; efficient, wavelet-based data storage; automated image registration based on mutual information and histogram transformations; and texture-based volume rendering for interactive rendering on multiple scales. Efficient storage and processing of multimodal images as well as histogram transformation and registration will be discussed. It will be shown how the conflict of variable resolutions that occurs when using different modalities can be resolved efficiently by using a wavelet-based storage pattern, which also offers advantages for multi-resolution rendering. [ABSTRACT FROM AUTHOR]

There are many application areas where dynamic visualization techniques cannot be used and the user can only view a still image. Perceiving depth and understanding spatio-temporal relations from a single still image are challenging tasks. We present visualization techniques which support the user in perceiving depth information from 3D angiography images, and techniques which depict motion inherent in time-varying medical volume datasets. In both cases no dynamic visualization is required. [ABSTRACT FROM AUTHOR]

Three integral components to build a tutoring system for echocardiography are presented. A mathematical time-varying model for vessel-representations of the human heart, based on cubic B-Splines and wavelets facilitating the extraction of arbitrarily detailed anatomical boundaries. A dedicated ontology framework the model is embedded into enabling efficient (meta-)data management as well as the automatic generation of (e.g. pathologic) heart instances based on standardized cardiac findings. A simulator generating virtual ultrasound images from instances of the heart transformed into isotropic tissue representations. [ABSTRACT FROM AUTHOR]

In this paper we present the framework of our reconstruction and visualization system for planar cross sectional data. Three-dimensional reconstructions are used to analyze the patterns and functions of dying (apoptotic) and dividing (mitotic) cells in the early developing nervous system. Reconstructions are built-up from high resolution scanned, routinely stained histological serial sections (section thickness = 1 μm), which provide optimal conditions to identify individual cellular events in complete embryos. We propose new sophisticated filter algorithms to preprocess images for subsequent contour detection. Fast active contour methods with enhanced interaction functionality and a new memory saving approach can be applied on the pre-filtered images in order to semiautomatically extract inner contours of the embryonic brain and outer contours of the surface ectoderm. We present a novel heuristic reconstruction algorithm, which is based on contour and chain matching, and which was designed to provide good results very fast in the majority of cases. Special cases are solved by additional interaction. After optional postprocessing steps, surfaces of the embryo as well as cellular events are simultaneously visualized. [ABSTRACT FROM AUTHOR]

Endoscopy is a popular procedure which helps surgeons investigate the interior of a patient's organ and find abnormalities (e.g., polyps). However, it requires a great expertise using only a stream of 2D images of the interior, and there is a possibility that the physician will miss some polyps. Instead, a 3D reconstruction of the interior surface of the organ will be very helpful. It turns the stream of 2D images into a meaningful 3D model. The physicians could then spend more time scrutinizing the interior surface. In addition, the 3D reconstruction result will provides more details about the patient's organ (e.g., concavity/convexity, a coordinate system and 3D measurements), and could be saved for later uses. In a related work, Helferty et al. [HZMH01, HH02] have used a CT-based virtual endoscopic registration technique to guide the bronchoscopic needle biopsy. Winter et al. [WSRW05] also proposed a reconstruction scheme enhanced by a data-driven filtering and a knowledge driven extension. [ABSTRACT FROM AUTHOR]

A large variety of techniques has been developed to visualize vascular structures. These techniques differ in the necessary preprocessing effort, in the computational effort to create the visualizations, in the accuracy with respect to the underlying image data and in the visual quality of the result. In this overview, we compare 3D visualization methods and discuss their applicability for diagnosis, therapy planning and educational purposes. We consider direct volume rendering as well as surface rendering. In particular, we distinguish model-based approaches, which rely on model assumptions to create "idealized" easy-to-interpret visualizations and model-free approaches, which represent the data more faithfully. Furthermore, we discuss interaction techniques to explore vascular structures and illustrative techniques which map additional information on a vascular tree, such as the distance to a tumor. Finally, navigation within vascular trees (virtual angioscopy) is discussed. Despite the diversity and number of existing methods, there is still a demand for future research which is also discussed. [ABSTRACT FROM AUTHOR]

Clinical radiologists require not only visual representations of MRI and CT data but also quantitative measurements representing the progression of chronic conditions such as rheumatoid arthritis of the knee. Since inflammation is confined to a thin irregularly shaped region called the synovial capsule it is necessary to segment a suitable approximation of the capsule, then compute quantitative measurements within the segmented region. We report preliminary results on applying topological tools to identify the desired region visually and to extract quantitative information, along with a protocol for clinical validation of the method. [ABSTRACT FROM AUTHOR]

We present a set of coherent methods for the nearly automatic creation of 3D geometric models from large stacks of images of histological sections. Three-dimensional surface models facilitate the visual analysis of 3D anatomy. They also form a basis for standardized anatomical atlases that allow researchers to integrate, accumulate and associate heterogeneous experimental information, like functional or gene-expression data, with spatial or even spatio-temporal reference. Models are created by performing the following steps: image stitching, slice alignment, elastic registration, image segmentation and surface reconstruction. The proposed methods are to a large extent automatic and robust against inevitably occurring imaging artifacts. The option of interactive control at most stages of the modeling process complements automatic methods. [ABSTRACT FROM AUTHOR]