Inhomogeneous deformation as a source of error in strain measurements derived from implanted markers in the canine left ventricle

This article quantifies the errors inherent in the measurement of myocardial strain in the canine left ventricle when the motion of four radiopaque marker beads is used to determine this strain. These errors are introduced because the strain is strongly inhomogeneous and only an averaged value of this strain can be determined by measuring the displacements of four points with finite separation. In this work, the error in the principal strains has been estimated by modeling the primary deformation components of the left ventricle and comparing the true strains obtained from these models with the strains computed according to the protocol typically used in experimental studies to determine strain from the motion of marker beads. Both a cylindrical and a spherical model of the left ventricle are used. For the cylindrical model, it is found that the traditional tetrahedra used may give errors as high as 20% in the maximum principal strain. A six-marker prism is found to give more consistent results, underestimating the maximum principal strain, which is in the radial direction, by no more than 8% in almost all cases. The spherical model, having double curvature, gives larger errors. In both models, the error in the other two principal strains was usually less than 5%. Furthermore, the principal strain directions were correct to within 6 degrees.