VMLS: Visualization in Medicine and Life Sciences

Permanent URI for this community

https://diglib.eg.org/handle/10.2312/1011

Browse

Now showing 1 - 11 of 11

Amyloid Analyzer - A software assistant for qualitative and quantitative analysis of Florbetaben PET scans
(The Eurographics Association, 2013) Weiler, F.; Dicken, V.; Strehlow, J.; Geisler, B.; Scarpa, M.; Pessel, M.; Stephens, A.; Hahn, H. K.; L. Linsen and H. -C. Hege and B. Hamann
Amyloid imaging is currently on the verge of becoming a vital imaging biomarker for the diagnosis and progressmonitoring of Alzheimer's disease. It is a positron emission tomography (PET) imaging technique based on tracers binding to b-amyloid plaques in the brain. These plaques are known to accumulate over time in the gray matter of the brains of AD patients. Images acquired with an amyloid binding tracer can be difficult to interpret, especially for cases showing an early stage of the disease. Also, precise quantification is challenging, because the cortical gray matter can not be well delineated from the images. In this work, we present a software assistant targeted at both qualitative and quantitative analysis of amyloid PET scans. It has been designed with the aim to be easy to use and integrate well into clinical workflows, while at the same time providing solid quantitative results for use e.g. in pharmaceutical trials.
Combined Volume Registration and Visualization
(The Eurographics Association, 2013) Capps, Arlie G.; Zawadzki, Robert J.; Werner, John S.; Hamann, Bernd; L. Linsen and H. -C. Hege and B. Hamann
We describe a method for combining and visualizing a set of overlapping volume images with high resolution but limited spatial extent. Our system combines the calculation of a registration metric with ray casting for direct volume rendering into a combined operation performed on the graphics processing unit (GPU). The combined calculation reduces memory traffic, increases rendering frame rate, and makes possible interactive-speed, usersupervised, semi-automatic combination of many component volume images. For volumes that do not overlap any other imaged volume, the system uses contextual information provided in the form of an overall 2D background image to calculate a registration metric.
Corrected Uncertainty in Probabilistic Segmentation Using Local Statistics
(The Eurographics Association, 2013) Ristovski, Gordan; Hahn, Horst; Linsen, Lars; L. Linsen and H. -C. Hege and B. Hamann
Probabilistic segmentation algorithms compute for each voxel and each segment of a medical imaging data set a probability that the voxel belongs to the segment. These per-voxel probability vectors are commonly used to estimate uncertainties and produce respective visualizations. It can be observed that one obtains high uncertainties along the border of two adjacent tissues, even in case of high gradients. This is due to the partial volume effect (PVE). PVE, however, is not a source of uncertainty. In case of high-gradient borders, one can be very certain that respective voxels partially belong to one and partially to the other voxel. We correct this misconception by modeling PVE using local statistics within a probabilistic segmentation approach. As a result we obtain corrected uncertainties and we even have been able to significantly improve the probabilistic segmentation approach itself.
Functional Connectivity Glyphs for Brain Visualization
(The Eurographics Association, 2013) Böttger, J.; Schurade, R.; Margulies, D. S.; L. Linsen and H. -C. Hege and B. Hamann
The correlation of spontaneous brain activity, termed: functional connectivity, has become a valuable method in recent years for mapping brain organization. We present a novel approach to visualize functional connectivity that displays full connectivity between nodes on the cortical surface. Functional connectivity glyphs make it possible to visualize the entire functional connectome within a single image, thus enabling a detailed mapping of different cortical areas based on their connectional fingerprint.
Hierarchical Poisson-Disk Sampling for Fiber Stipples
(The Eurographics Association, 2013) Hlawitschka, Mario; Goldau, Mathias; Wiebel, Alexander; Heine, Christian; Scheuermann, Gerik; L. Linsen and H. -C. Hege and B. Hamann
To understand neural tracts of the brain, neuroscientists use visualizations of diffusion data. Fiber stippling - a technique inspired by illustrations - accommodates probabilistic tracts, main diffusion direction, and anatomical context in the same slice image. It uses stratified sampling to place stipples, but this can result in overlaps and undersampled areas that distort the perception of tract probability. Moreover, when changing slices in an interactive setting, resampling leads to visual noise that distracts from real changes in the data. In this paper, we propose to use Poisson-disk samplings to ensure adequate pattern perception inside slices and a hierarchy of samplings to ensure coherence among slices. We also port the algorithm to the GPU to achieve interactive frame rates. Our modifications are appreciated by neuroscientists, who can now investigate white-matter structures more confidently.
Improving Electrical Impedance Tomography Imaging of the Lung with Patient-specific 3D Models
(The Eurographics Association, 2013) Salz, P.; Reske, A.; Wrigge, H.; Scheuermann, G.; Hagen, H.; L. Linsen and H. -C. Hege and B. Hamann
Electrical Impedance Tomography (EIT) visualizes conductivity changes inside the thorax, which correlate with breathing and cardiac activity. While featuring high temporal resolution, no patient risks and bedside application, EIT has a very low spatial resolution, and its anatomical correspondence depends crucially on the choice of body model for image reconstruction. In contrast to the state of the art simplified or averaged 2D body models, we propose a workflow to generate patient-specific 3D models from Computed Tomography (CT) segmentations. This method acknowledges the 3D characteristics of EIT-induced currents in the body, while measurements are only performed in 2D. The workflow was designed in collaboration with medical experts such that its applicability in the clinical context becomes feasible. This is in contrast to most other works that only consider isolated algorithms and neglect the clinical demands and tasks. Our approach generates CT segmentations using another novel workflow based on interactive sketching, computes a tetrahedral multi-material mesh and creates a forward model with these results. The GREIT reconstruction algorithm is used to generate EIT images using the 3D model, while its parameters are tuned to the 3D properties of the mesh. We present results from two pigs, with three EIT datasets each, including mechanical ventilation, ventilation under the influence of lung injury, and ventilationfree regional perfusion analysis. We discovered three anatomical phenomena in the improved EIT images that could be visualized and explained using our workflow, while they caused some confusion in image interpretation using the state of the art techniques. These results, though not yet quantitatively measured, show the improved image quality and better anatomical significance, and stress the importance of accurate body models for EIT application in clinical research and patient treatment.
Interactive Visualization of Neuroanatomical Data for a Hands-On Multimedia Exhibit
(The Eurographics Association, 2013) Rieder, C.; Brachmann, C.; Hofmann, B.; Klein, J.; Köhn, A.; Ojdanic, D.; Schumann, C.; Weiler, F.; Hahn, H. K.; L. Linsen and H. -C. Hege and B. Hamann
Magnetic resonance imaging is a technique which is routinely used by neuroradiologists. Within the last decade, several techniques have been developed to visualize those MR images so that medical experts, and thus the patients, can benefit from it. However, very little work has been done to use neuroanatomical MR data for educational purposes and to bring the general public into closer contact with the scientific knowledge. In this paper, an interactive visualization of neuroanatomical data, which is controlled by a dedicated user input device, is presented for a novel neuroscience exhibit. State-of-the-art visualization methods are combined to facilitate easy perception of the complexity of the medical data. For that, fiber tubes and diffusion-weighted image overlays are integrated into a volume rendering of the brain. Ambient occlusion algorithms are utilized to calculate self-shadowing of the brain anatomy and the fiber tubes. Further, a physical model of the brain and a touch display are used as user input devices. The visibility of fiber bundles can be intuitively controlled by activating touch sensors, which have been inserted into the physical brain model at the corresponding functional areas.
Supervised Kernel Principal Component Analysis for Visual Sample-based Analysis of Gene Expression Data
(The Eurographics Association, 2013) Long, Tran Van; Linsen, Lars; L. Linsen and H. -C. Hege and B. Hamann
DNA microarray technology has enabled researchers to simultaneously investigate thousands of genes over hundreds of samples. Automatic classification of such data faces the challenge of dealing with smaller number of samples compared to a larger dimensionality. Dimension reduction techniques are often applied to overcome this. Recently, a number of supervised dimension reduction techniques have been developed. We present a novel supervised dimension reduction technique called supervised kernel principal component analysis and demonstrate its effectiveness for visual representation and visual analysis of gene expression data.
Towards a Structured Analysis of Quantitative Descriptors from Segmented Biological Image Data
(The Eurographics Association, 2013) Leitte, H.; Portl, J.; Röder, I. V.; Schröder, R. R.; Wacker, I.; L. Linsen and H. -C. Hege and B. Hamann
Topological and morphological descriptions of (sub-)cellular structures play a central role in the understanding of biological processes. Deriving such descriptions from image data, however, is a challenging task that has so far only been addressed for individual objects at a coarse resolution with small numbers of samples. For larger samples, the structured analysis is highly challenging as little a priori knowledge exists. In this paper, we address the design of a generic parameter space for segmented objects that forms the basis for subsequent structural analysis. We detail theoretical considerations, discuss the proposed model using examples from electron microscopy, and summarize lessons learned for subsequent implementation and analysis.
Towards Automatic Extraction of the Myocardium in Temporal MRI Using Object-based Segmentation
(The Eurographics Association, 2013) Chitiboi, Teodora; Hennemuth, Anja; Linsen, Lars; Hahn, Horst; L. Linsen and H. -C. Hege and B. Hamann
Visualizing the deformation of the myocardium over an entire heart cycle is essential for the assessment of local defects in the heart muscle. For this purpose, an automatic segmentation of the myocardium is a key requirement, which would then also form the basis for inspecting cardiac morphology, function and perfusion. In this paper we present an automatic object-based approach for segmenting the left ventricle (LV). The algorithm uses prior knowledge about the topology and geometry of the segmented structure, which is integrated in a region-based graph representation of the image. Our method was tested on cine MR sequences with promising results, and it also performed well when applied to perfusion data.
Visualization for Understanding Uncertainty in the Simulation of Myocardial Ischemia
(The Eurographics Association, 2013) Rosen, Paul; Burton, Brett; Potter, Kristin; Johnson, Chris R.; L. Linsen and H. -C. Hege and B. Hamann
We have created the Myocardial Uncertainty Viewer (muView or µView) tool for exploring data stemming from the forward simulation of cardiac ischemia. The simulation uses a collection of conductivity values to understand how ischemic regions effect the undamaged anisotropic heart tissue. The data resulting from the simulation is multivalued and volumetric and thus, for every data point, we have a collection of samples describing cardiac electrical properties. µView combines a suite of visual analysis methods to explore the area surrounding the ischemic zone and identify how perturbations of variables changes the propagation of their effects.

Browse

Browsing VMLS: Visualization in Medicine and Life Sciences by Title

Results Per Page

Sort Options