VISUALIZATION OF VASCULATURE FROM VOLUME DATA

In biomedical computing the need for visualization methods of the huma n vascular system has been triggered by recent advances in image acqui sition technology. In this paper we describe a number of approaches fo r the three-dimensional display of vessels from volumetric datasets. O ur approaches are based on the analysis of the deficiencies of the Max imum Intensity Projection algorithm, which today is the state-of-the-a rt technique for vascular display, and takes into account different di agnostic and therapeutic situations. For the qualitative as well as qu antitative evaluation of the major vessels, e.g.. the carotids, a mode l-driven Computer Vision method to segment, reconstruct and render the vascular tree surface including branches is presented. For the assess ment of smaller vessels a Pattern Recognition technique for contrast e nhancement of line-like structures is introduced. It serves as a prepr ocessing step prior to the application of a volume-rendering algorithm that consists of an advanced maximum projection scheme with depth-cue ing. The integrated 3D display of soft-tissue surfaces with adjacent v asculature, as required, e.g., for neurosurgery planning, is enabled b y a raycaster, simultaneously rendering and merging two volume dataset s. We emphasize the clinical relevance of techniques for explorative v olume data analysis by a walkthrough example. The paper demonstrates t he necessity of incorporating Computer Vision and Pattern Recognition methodologies into the Scientific Visualization pipeline for biomedica l imaging.