Application of a connected-voxel algorithm to MR angiographic data.

A connected-voxel algorithm (CVA) that improves the contrast and conspicuity of blood vessels in maximum-intensity-projection (MIP) magnetic resonance (MR) angiography is described. Images from a variety of anatomic regions in healthy volunteers were calculated with either an MIP procedure alone or with data that had first been processed with the CVA. A low-signal-intensity threshold is first applied to separate groups of voxels associated with different vessels from one another and to eliminate the contribution from low-intensity stationary material. The remaining voxels are grouped by a connectivity criterion into discrete "objects." Vessels are represented by extended objects, and small objects are discarded. The CVA, therefore, reduces the full three-dimensional data set into a small number of discrete objects. It is a powerful technique that can be used to remove signal from vessels overlying the vessel of interest, to separate objects representing arterial flow from those representing venous flow, to eliminate flow artifact from projection images, and to more completely retain signal within the vascular lumen. This technique has been successfully demonstrated with MR angiography in healthy volunteers.