Cardiac Strain Assessment for Fibrotic Myocardial Tissue Detection in Left Ventricular Hypertrophic Cardiomyopathy

This work proposes an image processing methodology to distinguish fibrotic from normal tissue by the assessment of the local mechanical properties of the myocardium in magnetic resonance tagging images. The procedure uses the information provided by short axis images of the above mentioned modality to estimate the Green-Lagrange strain tensor; a modified method based on the Harmonic Phase is employed for motion estimation. The method has been applied to the analysis of the local deformation patterns in a set of patients affected by hypertrophic cardiomyopathy in order to find the agreement between hyperenhanced zones in late enhancement images and areas in the myocardium with abnormal tensor values (both the radial and the circumferential components as well as the shearing component have been accounted for). The agreement is measured taken as ground truth manual segmentation of late enhancement images carried out by two cardiologists. Finally, a set of example images illustrate the agreement between both techniques.