MR Tag Surface Tracking Using a Spatio -

Magnetic resonance imaging provides the unique capability to produce artiicial, high-contrast features called \tags" that are invaluable for tracking tissue motion. Each tag corresponds to an initially planar surface embedded in the tissue that deforms with tissue motion. By tracking tag surface deformation, quantitative analysis of tissue motion can be performed. In this paper, we present a method for tag surface tracking that applies speciically to tag surfaces embedded in the wall of the left ventricle. The method addresses two key issues: rst, the full spatial extent of tag surfaces in 3-D space must be inferred from 2-D images, and second, within the images, noise leads to uncertainty in tag positions. We address these issues by framing tag surface tracking as an estimation problem given the observed image data. The estimates are obtained using a stochastic model of tag deformation and a recursive algorithm that simultaneously lters over time and smoothly interpolates between images. 1 Introduction Magnetic resonance tagging 1] is a technique used for tissue motion analysis, with applications primarily in the study of left ventricle (LV) function 2,3]. Simply stated, tags are non-invasively induced features that appear as dark \tag lines" in MR images. These features deform with the tissue, providing a visual motion cue in the images. A more physical depiction is provided by Figure 1, which shows that tags actually correspond to surfaces embedded in the tissue and each \tag surface" typically intersects several image planes, producing one tag line in each image. When applied, tag surfaces correspond to planes of tissue, but they are subsequently deformed by LV motion. Using tagged MRI, we previously presented a method for tracking the 3-D motion of large numbers Figure 1: Tag surface intersecting a stack of MR images. The bright ring is the LV wall and the dark lines are tags, each corresponding to a tag surface. of material points within the wall of the LV 3]. A key step in the method is tracking the deformation of tag surfaces by reconstructing them at regular time intervals. This is done by obtaining stacks of images, such as the one shown in Figure 1, at each time. Then, sets of points are identiied along tag lines in each image and complete surfaces are reconstructed by interpolating between points using thin-plate splines. However, this approach to tag surface reconstruction neglects the error inherent in identifying tag lines. Available spatial and …