Validation of tagging with MR imaging to estimate material deformation

Abstract

Myocardial tagging with magnetic resonance imaging is useful for non-invasive estimation of in vivo heart wall deformation. To validate the method of strain estimation and quantify the error of deformation estimates, a deformable silicone gel phantom in the shape of a cylindrical anulus was built and imaged. Four observers digitized the displacement of magnetic tags in two deformation modes: axial shear, caused by a 45 degrees rotation of the inner cylinder, and azimuthal shear, caused by a 13.5-mm longitudinal translation of the inner cylinder. In axial shear, good agreement was found between the angular displacement of stripes painted on the gel and an analytic solution. Displacement of magnetic tags also agreed with that solution. Interobserver and observer-model errors in deformation estimates were quantified for homogeneous and nonhomogeneous strain analysis. In homogeneous strain analysis, errors in point localization produced relatively large errors, which were reduced in nonhomogeneous strain analysis. Both estimates were unbiased across the range of deformations.