Authors: I. Viola, A. Kanitsar, M. Gr ö ler Institute of Computer Graphics and Algorithms Vienna University of Technology, Austria Importance Driven Volume Rendering
Introduction (problem) Volume visualization has inherent difficulties with occlusion Exterior objects occlude interior objects Interior objects might be more interesting or important
Introduction (motivation) 3D medical image visualization in general Specific example: liver tumor visualization For radiotherapy planning For surgery planning Need to clearly see: Tumor location, size, shape Blood vessel tree Parenchyma
Introduction (solution) object importance level of sparseness make interesting objects occluded clearly visible Artistic medical illustration Method proposed in this paper
Related Work Transfer Function Incorporate first and second derivatives (gradient and curvature values) Degree of interest (Hauser & Mlejnek) Focus +Context Rendering Focus: Render that part with more detail, with magnification, or other enhancement Context: according to Distance to focal point, d etail or magnification gradually fades as distance increases
Related Work Sparse Representation Interior structures — fully opaque Enclosing objects — curvature-directed lines Cut-Away Views
Importance Driven Rendering Importance is widely used in other graphics fields Employ as additional dimension to improve the behavior of traditional approaches Apply this in volume rendering
Importance Driven Volume Rendering Object Importance Importance Compositing Levels of sparseness
Object Importance Additional property besides color and opacity Positive scalar value to show priority of each object importance Constant for the whole object
Levels of sparseness Various different representations of a certain object from dense to sparse
Importance Compositing Specify the level of sparseness Simple way:Maximum importance projection (MImP) Average Importance Compositing
Importance Compositing Mode 1: Maximum Importance Projection highest importance—visible Highest sparseness—transparent Assign sparseness level of 0 or 1 to each point 0 1 1
Problems with MImP The spatial management of structures is not readily apparent.
Improved MImP Consider MImP as a cut-away view Cylindrical Conical Countersink
Mode 2: Average Importance Compositing Compute sum of object importances along a ray Each surface point intersected by the ray is assigned a sparseness level based on the ratio of its importance to the sum of all intersected surface importances Importance Compositing 0 1 2
Smooth Transition with level of sparseness With Avg. Importance Compositing, generate jaggy object boundaries Color and Opacity Modulation Screen-Door Transparency Volume Thinning
Use transfer function to modulation the saturation of color and opacity Color and Opacity Modulation
Screen-Door Transparency Semi-transparency Object is Visible through the holes Mesh occludes the object behind
Volume Thinning Render objects using a set of iso- surfaces
Results Pre-segmented objects Leopard Gecko 512*512*87 Monster Study 256*256*610
Conclusion A new factor to traditional volume rendering—object importance Apply level of sparseness if interesting structures are occlude Methods to smooth transitions
Future work How to automatically assign importance Figure out the application in Volume Graphics Incorporate 1 st and 2 nd order derivatives to preserve object boundaries in composition schemes How to automatically determine optimal viewpoints