Cardiac Strain Imaging with Dynamically Skipped Frames: A Simulation Study

Accurate lateral displacement estimation is a critical requirement to reliably use cardiac strain imaging for myocardial health or carotid strain imaging to assess atherosclerosis. Estimation of small lateral displacements are challenging even with 2-D kernels. We investigate the feasibility of using a dynamic frame skip (DFS) approach to improve lateral strain estimation for cardiac and carotid strain imaging. DFS automatically determines the optimal frame skip using axial strain information estimated during displacement tracking. We incorporated DFS into a hierarchical Bayesian regularization based displacement tracking method for cardiac and carotid strain imaging. A finite element analysis (FEA) model of canine cardiac deformation was utilized to quantify estimation performance with DFS and compare against conventional inter-frame displacement (CFS) approach. DFS displacement and strain temporal curves showed improved qualitative agreement with FEA ground truth compared to CFS results. Consequently, DFS had lower strain total temporal relative (TTR) strain errors (TTRaxial= 18.37 %, TTRlateral= 29.91 %, TTRradial = 17.32 %, and TTRlongitudinal = 14.51 %) compared to CFS (TTRaxial= 22.39 %, TTRlateral = 30.88 %, TTRradial = 22.35 %, and TTRlongitudinal= 15.87 %). These results show that DFS can potentially improve cardiac strain estimation quality.