The Dilation-shadowing Approach for Volume Registration on Small Clusters

T.S. Newman, X. Zhang, and J.D. Bush (USA)


medical imaging, data treatment and visualization, volume warping, volume registration


The Dilation-Shadowing, a new approach for aligning volume datasets with curve-like features, is introduced. In addition, this paper presents a load-balanced scheme for efficient application of the approach on clusters with a small number of nodes. The dilation-shadowing is motivated by a shadow generation technique from the computer graphics and achieves its efficiency through an inverse approach that is based on morphological dilation. The method has been applied to the warping of magnetic resonance angiography (MRA) data of the brain and uses the change in positions of curved structures (i.e., blood vessels, which are easily extracted from MRA data) to guide the warping process. The warping method, details of the load balancing scheme, and performance results from a realization on a cluster computer are presented.

Important Links:

Go Back