Julian Kreiser, Forschungsgruppe Visual Computing, stellt sein Dissertationsvorhaben vor.
Medical data visualization, as a subfield of scientific data visualization, concerns itself mainly with spatiotemporal data sets. However, with the ever-increasing amount and resolution of such data sets, medical experts face the difficulty of interpreting this plentitude of information. With this trend in mind, diagnosis and treatment duration, accuracy, and the patients safety are important factors when designing medical visualizations. Naturally, these challenges lead to the question, weather the amount of presented data can be minimized while preserving relevant features to avoid a cognitive overload of the medical expert. As context-aware projection techniques can be employed to allow for such a workflow optimization, in this dissertation, projection-based techniques are explored and developed within the medical visualization domain. The overall goal is to assist experts in fulfilling their tasks more effectively.
The thesis begins with a literature analysis to investigate attribute preserving flattening techniques in the domain of medical visualization. These methods are often used for subsequent algorithmic analysis and
processing tasks to extract further information from a single patient or over multiple subjects.
The following novel approaches for different applications demonstrate the value of projection as a core concept to communicate the maximum information through task-specific visualization. The conducted research aims to aid medical experts in diagnostic and intra-operative tasks.
First, a navigational support visualization has been developed for irreversible electroporation interventions for tumor ablation. The difficulty during such a procedure is to not damage any critical structures around the target area to ensure a patient's safety and fast recovery. To support surgeons, the visualization helps to precisely place multiple ablation needles whose position and orientation are dependent on each other.
Second, an abstract visualization for high-resolution manometry imaging data facilitates the diagnosis of esophageal mortility disorders. The goal of this work is to represent a patient's spatio-temporal pressure data of several test swallows in a single, visual descision graph. Typically, each swallow is inspected independently. This makes it difficult to get a good overview of the whole data set, needed for diagnosis. The developed embedding allows experts now to form a fast and accurate diagnosis using a single view.
Third, a novel approach, called Void Space Surfaces, improves the depth perception of vascular geometry.
The empty space between vessels is used to generate a height field. This surface then serves as a canvas to convey depth without having to interfere with other visual encodings on the vascular structures itself.
Bei Interesse melden Sie sich bitte beim Institut für Medieninformatik, Frau Duschek.