Fast techniques for determining myocardial strains from tagged MRI

Summary form only received. Magnetic resonance (MR) tagging is a useful technique for noninvasively measuring the deformation of the heart. With MR tagging, quantitative measurement of heart motion can be obtained from the motion of a set of dark lines encoded within the deforming Left-Ventricular (LV) tissue. However, these tag lines are sparse and therefore techniques are needed to reconstruct a dense displacement field at all points within the LV myocardium so that mechanical indices of function such as strain, shear, and torsion can be determined for all myocardial points. Here, a subspace approximation technique is used to accomplish this task. First, the authors formulate the displacement estimation as a variational problem and then a subspace approximation method which makes use of B-spline bases is proposed to approximate the solution. By taking advantages of B-spline properties, efficient numerical methods are derived which reconstruct the motion field within a short-axis MRI slice in under one minute. The methods have been extensively validated using both simulated and in-vivo data in animal models and patients.