Ken C. L. Wong PhD, Abstract Co-Author: Nothing to Disclose

Michael Tee BS, Presenter: Nothing to Disclose

Marcus Yen-Ta Chen MD, Abstract Co-Author: Institutional research agreement, Toshiba Corporation

J. Alison Noble, Abstract Co-Author: Nothing to Disclose

David A. Bluemke MD, PhD, Abstract Co-Author: Research support, Siemens AG

Ronald M. Summers MD, PhD, Abstract Co-Author: Royalties, iCAD, Inc Research funded, iCAD, Inc Stockholder, Johnson & Johnson Grant, Viatronix, Inc

Jianhua Yao PhD, Abstract Co-Author: Royalties, iCAD, Inc

Regional myocardial strains have the potential for earlier quantification and detection of cardiac dysfunction compared with global geometric measurements. Although image modalities such as tagged and CSENC MRI can provide motion information of the myocardium, they are not common practices in clinical routine. Therefore, accurate strain estimation from more available cardiac CT can be beneficial.

As cardiac CT can only provide motion information of salient features such as heart surfaces, additional information is required. For physiologically plausible and clinically applicable strain estimation, we propose to use a hyperelastic biomechanical model. Four canine cardiac CT sequences with artificially induced myocardial infarction were used for evaluation. Each cardiac cycle (0.52-0.89s) has 20 frames with voxel size < 0.3x0.3x1 mm3. The infarcted regions were identified by experts using perfusion CT on the American Heart Association nomenclature (17 zones). Image segmentation and meshing were performed at end-diastole to provide the finite element (FE) model of the heart. Deformable image registrations based on B-spline interpolation and mutual information were performed on the whole sequence to provide the motion information on the heart surfaces. The displacements between frames were enforced as boundary conditions to the FE model, and hyperelastic biomechanics was applied to compute the quantitative myocardial strains.

The estimated zonal first principal strains at end-systole are consistent with the infarction manually identified by experts, with the average Cohen’s kappa as 0.77+/-0.22. The average strain value of the 22 infarcted zones is 0.13+/-0.07, and that of the 46 normal zones is 0.37+/-0.17 (p<0.0001).

With the use of the hyperelastic biomechanical model, myocardial strains can be quantified from the motion of the salient cardiac features extracted from images. The consistency between the expert identification and the estimated strains shows that this framework is promising for cardiac diagnosis.

Cardiac CT is commonly available in clinical routine and the regional myocardial strains estimated from CT have the potential for early quantification and detection of cardiac dysfunction.

Wong, K, Tee, M, Chen, M, Noble, J, Bluemke, D, Summers, R, Yao, J, Regional Myocardial Strain Estimation with Hyperelastic Biomechanical Model: Application on Cardiac CT.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14014771.html