Visualization Techniques for Virtual Endoscopy
(Vilanova i Bartrolí, Oct 2001) Vilanova i Bartrolí, Anna
Virtuelle Endoskopie ist eine Technik hohle Organe und anatomische Aush¨ohlungen mittels Methoden der medizinischen 3D Visualisierung und Verfahren aus der Computergraphik zu erforschen. Virtuelle Endoskopie ist eine vielversprechende Technik, welche die tats¨achliche Endoskopie verbessert, oder in manchen Anwendungen sogar ersetzen kann. Endoskopie ist eine invasive und f¨ur den Patienten risikoreiche Methode. In den letzten Jahrzehnten erfolgte viel Forschungsarbeit in diesem Bereich. Diese Dissertation reflektiert einen Teil dieser Forschung und im speziellen die Untersuchung von Visualisierungstechniken f¨ur die virtuelle Endoskopie. Die meisten Methoden der virtuellen Endoskopie simulieren das Verhalten echter Endoskopie. Der erste Teil der Dissertation konzentriert sich auf die Betrachtung dieser Methoden und stellt die Struktur eines solchen Systems vor. Ein Prototyp angelehnt an diese Struktur wird entwickelt. Ein zentrierter Pfad innerhalb der Organe wird verwendet, um eine Kamera durch das Organ zu f¨uhren. Wir verbessern einen existierenden Algorithmus, um den zentrierten Pfad durch das Organ zu finden. Zwei neue Techniken werden beschrieben, um hochqualitative, perspektivische Darstellungen zu beschleunigen. Die erste Methode ist eine neue Space-Leaping-Technik f¨ur Volumsdarstellungen. Die zweite Methode n¨utzt Hardware-Beschleunigung (VolumePro) zur orthographische Volumsdarstellung, um perspektivische Volumsdarstellungen zu erzeugen. Simulationen von wirklicher Endoskopie ist in vielen Anwendungen nicht die am besten geeignete Visualisierungs-technik. Eine Endoskopie ist an gewisse physikalische Beschr¨ankungen gebunden, welche f¨ur die virtuelle Endoskopie nicht gelten. Der zweite Teil der Dissertation pr¨asentiert zwei Techniken, welche den Dickdarm virtuell entfalten, um die Oberfl¨ache zu untersuchen, und Polypen entdecken zu k¨onnen. Wir konzentrieren uns auf den Dickdarm, obwohl diese Techniken auch f¨ur andere Organe einsetzbar sind. Die erste Technik entfaltet den Dickdarm lokal und erzeugt eine animierte Abfolge von aufeinanderfolgenden entfalteten Regionen. Die Bilder werden mittels einer Projektionstechnik erzeugt, welche dem Mediziner die Visualisierung eines Großteils der Oberfl¨ache erlauben. Diese Methode erlaubt es Polypen einfach zu erkennen, welche aus dem Blickwinkel einer Endoskopie von Falten verdeckt, oder schwer zu lokalisieren w¨aren. Der Nachteil der lokalen Dickdarmentfaltung ist die Notwendigkeit die Untersuchung anhand eines Films durchf¨uhren zu m¨ussen, um die gesamte Oberfl¨ache zu visualisieren. Die zweite Technik Nonlinear Colon Unfolding erlaubt dem Mediziner in einer einzelnen Darstellung die Oberfl¨ache des Organs zu untersuchen und m¨oglichst viel Information darzustellen. Man erh¨alt ein einzelnes Bild des vollst¨andig entfalteten Dickdarms. Auf diese Weise k¨onnen problembehaftete Bereiche schnell identifiziert und anschließend genauer untersucht werden. Um die Anwendbarkeit zu zeigen wurde jede Technik wurde an mehreren Datens¨atzen getestet. Jedoch sind diese Tests nicht ausreichend, um die Methoden bereits in einem klinischen Umfeld einsetzen zu k¨onnen. Die Resultate zeigen jedoch das Potential der entwickelten Methoden auf. - Virtual Endoscopy is a technique to explore hollow organs and anatomical cavities using 3D medical imaging and computer graphics. Virtual Endoscopy turns out to be a promising technique to improve, or even in some procedures substitute, real endoscopy. A real endoscopy is invasive and usually implies some risk for the patient. In the last decades, much research has been done in this field. This thesis reflects a piece of this research concentrated on investigating visualization techniques for virtual endoscopy. Most common methods in virtual endoscopy simulate the behavior of a real endoscope. In the first part of this thesis, we concentrate on studying these methods and present a general framework for such a system. We develop a prototype according to this framework. A central path of the organ is used to move the camera through the organ. We improve an existing algorithm to find the central path of the organ. We describe two new techniques to accelerate high quality perspective volume rendering. The first method is a new space leaping acceleration technique for ray casting. The second method uses hardware acceleration (i.e., VolumePro) of orthographic volume rendering to generate perspective volume rendering. Simulating a real endoscopy is not the most suitable visualization technique in many endoscopy procedures. A real endoscopy is restricted due to physical limitations that a virtual endoscopy does not have. The second part of the thesis presents two techniques which virtually unfold the colon to inspect its surface and detect polyps. We concentrate on the colon although these techniques could be used with other organs too. The first technique, locally unfolds the colon and generates an animation sequence from consecutive unfolded regions. The images are generated with a projection technique that allows the physician to visualize most of the surface, and to easily recognize polyps that in an endoscopic view would be hidden by folds or would be hard to localize. The drawback of the local colon unfolding is that the physician has to inspect a video to be able to visualize the whole surface. The goal of the second technique, nonlinear colon unfolding, is to enable the physician to inspect and get as much information as possible of the inner organ surface at a first glance. It obtains a single image of the complete unfolded colon. In this way, the problematic areas can be identified quickly and inspected later in more detail. Every technique have been tested with several data sets to show their feasibility. Although, these tests are not enough to use the methods in the clinical environment, the results show their potential to achieve such an state.
Graphical Abstraction and Progressive Transmission in Internet-based 3DGeoinformationsystems
(Volker Coors, 2003) Volker Coors
The aim of this thesis was to eliminate essential deficits in the use of previous 3D GIS in an openly distributed GIS infrastructure. The increasing availability of three-dimensional city models poses a special problem, because previous GIS were not able to use such a database sensibly. An additional difficulty occurs in computer graphical representations, where problems exist when generating a meaningful, interactive, three-dimensional presentation of these large models in a heterogeneous network environment. The use of three-dimensional geodata in new applications like Location Based Services is only feasible when these basic technical problems are solved. This chapter recapitulates the results of the thesis and their significance for GIS research and summarizes the applications. The perspective on future work in this research field is developed subsequently.
Derivatives and Eigensystems for Volume-Data Analysis and Visualization
(Hladuvka, Dec 2001) Hladuvka, Jiri
Der Begriff Volumsdaten beschreibt diskrete drei-dimensionale r¨aumlichen Signale. Verfahren, welche auf diesen Daten arbeiten, k¨onnen grob in zwei Kategorien unterteilt werden: Visuelle Verfahren, welche das Ziel haben, eine f¨ur den Menschen interpretierbare Ausgabe zu liefern. Analytische Verfahren bereiten die Daten f¨ur weitere maschinelle Bearbeitungen auf. Obwohl beide Disziplinen (Volumsvisualisierung und Volumsverarbeitung) große ¨Ahnlichkeiten aufweisen, werden Themenstellungen oft getrennt behandelt. Diese Dissertation behandelt die Klassifikation von Volumsdaten basierend auf der Beobachtung von Unterschieden der skalaren Werte in einer lokalen Umgebung. Ein ¨ubliche Ansatz solcher Untersuchungen ist es, die Taylor-Entwicklung skalare Datens¨atze zu untersuchen. Als ein wichtiges Thema stellt sich bei der Analyse in h¨oheren Dimensionen die zu untersuchenden Richtungen heraus. Gewisse Antworten k¨onnen Eigensysteme von algebraischen Strukturen liefern, welche Ableitungen erster oder zweiter Ordnung beinhalten. Nach einem kurzen ¨Uberblick ¨uber, auf Ableitungen basierenden Klassifizierungen in der Volumsvisualisierung und Volumsverarbeitung, pr¨asentieren wir neue Beitr¨age, welche auf drei verschiedene Probleme anwendbar sind: Spezifizierung von Transferfunktionen, Bestimmung der relevanten Merkmale im Volumendatensatz und merkmalgesteuerte Rekonstruktion durch Interpolation. - Volume data refer to sampled three-dimensional spatial signals. The tools which handle them can broadly be divided into two categories: visual tools which aim at an output interpretable by a human user and analytic tools which prepare the data for further machine processing. Although it would be natural that the two related disciplines, i.e., volume visualization and volume processing closely collaborate, they are still rather separated. The work presented here contributes to bridge the gap in between. This thesis addresses classification of volume samples based on observations of how the scalar values vary in their vicinity. We investigate the first three terms of a Taylor series expansion of the corresponding scalar field at the inspected points. An important issue arising with such an analysis in higher dimensions are the directions to be examined. In order to find an answer to this problem we study the eigensystems of algebraic structures composed of the first-and second-order partial derivatives. After a survey on derivative-based classification in volume visualization and processing we present new contributions which apply to three distinct problems: specification of transfer functions, content-based retrieval of volume-data features, and shape-based interpolation.
Efficient, Image-Based Appearance Acquisitionof Real-World Objects
(Lensch, Hendrik Peter Asmus, 2003-12-15) Lensch, Hendrik Peter Asmus
Two ingredients are necessary to synthesize realistic images: an accurate rendering algorithm and, equally important, high-quality models in terms of geometry and reflection properties. In this dissertation we focus on capturing the appearance of real world objects. The acquired model must represent both the geometry and the reflection properties of the object in order to create new views of the object with novel illumination. Starting from scanned 3D geometry, we measure the reflection properties (BRDF) of the object from images taken under known viewing and lighting conditions. The BRDF measurement require only a small number of input images and is made even more efficient by a view planning algorithm. In particular, we propose algorithms for efficient image-to-geometry registration, and an image-based measurement technique to reconstruct spatially varying materials from a sparse set of images using a point light source. Moreover, we present a view planning algorithm that calculates camera and light source positions for optimal quality and efficiency of the measurement process. Relightable models of real-world objects are requested in various fields such as movie production, e-commerce, digital libraries, and virtual heritage.
Alves dos Santos, Luiz: Asymmetric and Adaptive Conference Systems for Enabling Computer-Supported Mobile Activities
(Alves dos Santos, Luiz Manoel, 2003) Alves dos Santos, Luiz Manoel
This work was conducted at the Darmstadt University of Technology, essentially between 1998 and 2002. Before and during this period, I was working at the INI-GraphicsNet, Darmstadt, first in the Zentrum für Graphische Datenverarbeitung e.V., and then later at the Fraunhofer-Institut für Graphische Datenverarbeitung (IGD), as a researcher. This thesis addresses the investigations and results achieved during my work at these organizations. My initial development projects in the area of mobile computing were very challenging due to the immense constraints posed by the then incipient hardware and wireless network infrastructures, and similarly overwhelming due to the desire to employ those fascinating appliances by all means possible. The endeavour to keep the respective application systems in a course of continuous improvement (i.e., with richer media presentation and “interactiveness”), and at the same astonishing pace as the technological evolutions, was both demanding and rewarding; however, it turned out to be a questionable procedure. After several prototype demonstrations and observations, there came a turning point, following the acknowledgement that, for application cases involving user mobility, the supporting tool is appraised significantly on the basis of its adequacy for the usage conditions and its flexibility to adapt to changing requirements and to any platform specification or resource availability. The circumstances of a mobile use (e.g., outdoor, on the move, in confined places) require new approaches in application system development and create a high demand for specialized, task-oriented system features. Any service being offered has to be able to account for, adjust itself, and be responsive to the increasing and unpredictable diversity of prospective users and their usage environments. The achievement of this attribute is even more challenging when the service should be a basis for a digitally mediated human-to-human communication process involving all kinds of diversity between the individual partners and technical arrangements. In this thesis work, proposals and innovative solutions to these challenges have been investigated and implemented, and are presented in this report. Some contributions of this work are: an adaptive conference system for heterogeneous environments, tools to assess, distribute, and respond to User Profiles at both the individual and collective level; adaptive, flexible individual interaction modes and media that are nevertheless consistent for a collaborative work; and mechanisms for remote awareness (of constraints) for structuring interaction. However, above any technological advances, the major research challenge was concerned with the human factor and the achievement of an effective integration of a supporting tool in their daily activities and lives.
Efficient Acquisition, Representation, and Rendering of Light Fields
(Schirmacher, Hartmut, 2003-12-16) Schirmacher, Hartmut
In this thesis we discuss the representation of three-dimensional scenes using image data (image-based rendering), and more precisely the so-called light field approach. We start with an up-to-date survey on previous work in this young field of research. Then we propose a light field representation based on image data and additional per-pixel depth values. This enables us to reconstruct arbitrary views of the scene in an efficient way and with high quality. Furthermore, we can use the same representation to determine optimal reference views during the acquisition of a light field. We further present the so-called free form parameterization, which allows for a relatively free placement of reference views. Finally, we demonstrate a prototype of the Lumi-Shelf system, which acquires, transmits, and renders the light field of a dynamic scene at multiple frames per second.
A Head Model with Anatomical Structure for Facial Modeling and Animation
(2003) Kähler, Kolja
In this dissertation, I describe a virtual head model with anatomical structure. The model is animated in a physics-based manner by use of muscle contractions that in turn cause skin deformations; the simulation is ef cient enough to achieve real-time frame rates on current PC hardware. Construction of head models is eased in my approach by deriving new models from a prototype, employing a deformation method that reshapes the complete virtual head structure. Without additional modeling tasks, this results in an immediately animatable model. The general deformation method allows for several applications such as adaptation to individual scan data for creation of animated head models of real persons. The basis for the deformation method is a set of facial feature points, which leads to other interesting uses when this set is chosen according to an anthropometric standard set of facial landmarks: I present algorithms for simulation of human head growth and reconstruction of a face from a skull.

Browse

Recent Submissions

Results Per Page

Sort Options