In communicating clinical impressions of vascular abnormalities between radiologists and physicians, it is often desirable to have a single image that visualizes all clinically relevant branches at once. The work presented in this dissertation is a development of methods for visualizing the abdominal aorta and its branches, whose tomographic images might be obtained by computed tomography (CT) or magnetic resonance (MR) angiography, in a single two-dimensional stylistic image, without overlap among branches. I call this two-dimensional visualization uncluttered single-image visualization (USIV).;This dissertation presents two methods for USIV that differ mainly in their global optimization strategies. The first method adopts simulated annealing to solve the optimization problem. The second method utilizes an interesting connection of the optimization problem regarding USIV to the protein structure prediction problem, in the context of global optimization. I performed a preliminary evaluation of the first method by asking three radiologists to label 366 arterial branches from the 30 visualizations of five cases produced by our visualization method. Each of the five patients was presented in six different variant images, selected from 10 variants with the three lowest and three highest scores. For each label, the radiologists assigned confidence and distortion ratings (low/medium/high). I studied the association between the quantitative metrics measured from the visualization and the subjective ratings by the radiologists. All resulting visualizations were free of branch overlaps. The labeling accuracy of the three readers was 93.4%, 94.5%, and 95.4%, respectively. For the total 1098 samples, the distortion ratings were low: 77.39%, medium: 10.48%, high: 12.12%, and the confidence ratings were low: 5.56%, medium: 16.50%, high: 77.94%. The association study shows that the proposed quantitative metrics can predict a reader's subjective ratings, and suggests that the visualization with the lowest score should be selected for readers. A statistical comparison of the second method with the first shows that the two methods are similar in the quality of outputs, while the second runs much faster than the first.;These encouraging results of the quantitative and subjective evaluation of both methods for USIV demonstrate their promise for increasing the effectiveness of communicating clinical findings among practitioners.;In this dissertation I propose a geometric framework for USIV. The abdominal aortic vasculature is modeled as an articulated object whose underlying topology is a rooted tree. Each part of the articulated object model abstracts the corresponding vessel segment by enclosing it in a bounding box whose shape is determined by the centerline and the radii of the vessel segment along its length. Then a method for USIV is formulated as a global optimization problem where the goal is finding a spatial configuration of the bounding boxes most similar to the projection of the input along a given viewing direction (e.g., the anteroposterior), while not introducing intersections among the boxes. The method defines and minimizes a score function that evaluates the overlap of the bounding boxes and the deviation of their configuration from the reference configuration. The output of the method is used to produce a stylistic visualization, made of flattened centerlines modulated by the associated radii information, on a plane.
展开▼
