Volume Rendering of Segmented Image Objects

This paper describes a new method of combining ray-casting with segmentation. Volume rendering is performed at interactive rates on standard laptop computers, and visualizations include both "superficial" ray-casting through a shell at each object's surface and "deep" ray-casting through the confines of each object. A feature of the approach is the option to smoothly and interactively dilate segmentation boundaries along all axes. This ability, when combined with selective "turning off" of extraneous image objects, can help clinicians detect and evaluate segmentation errors that may affect surgical planning. We describe both a method optimized for displaying tubular objects and a more general method applicable to objects of arbitrary geometry. In both cases, select 3D points are projected onto a modified z-buffer that records additional information about the projected objects. A subsequent step selectively volume renders only through the object volume...

• Type: Journal
• Year: 2002
• Where: TMI
• Authors: Elizabeth Bullitt, Stephen R. Aylward
• Keywords: Ray-casting, Segmentation Errors, Standard Laptop Computers, TMI 2002