To use three-dimensional displacement-encoded MRI and mesh-free strain analysis to characterize biventricular function in healthy subjects with multiple strain and timing parameters.Advanced measures of cardiac mechanics such as strain and torsion have been extensively characterized in the left ventricle (LV) and have been shown to predict patient outcomes.¹ However, the right ventricle (RV) is rarely studied and its contribution to cardiac function is not well understood. This knowledge gap is partly due to the technical challenges of imaging the thin wall, complex geometry, and irregular contraction pattern of the RV. Three-dimensional Displacement Encoding with Stimulated Echoes (DENSE) can now overcome these technical challenges. In this study, we have combined 3D DENSE imaging with custom post-processing that utilizes a local coordinate system to extract advanced measures of cardiac mechanics from both the right and left ventricles in an effort to characterize healthy biventricular function.

3D Spiral cine DENSE was performed on 40 healthy subjects (age: 27±8 years; 53% female) at 3T (Siemens Trio). Short-axis images were prescribed to cover both the right and left ventricles at end-diastole. Additional acquisition parameters included: 12 spiral interleaves, 360x360 mm2 FOV, 180x180 acquisition matrix, 8 mm slice thickness, TE/TR = 1.08/17 ms and 0.04 cycles/mm encoding frequency.^2,3 Three-point phase cycling was used for artifact suppression. All imaging was performed using a respiratory navigator.

RV and LV endocardial boundaries and an epicardial boundary were manually delineated on all cardiac phases using custom software (Figure 1A). The X, Y, and Z phase data within the myocardium were unwrapped using a quality-based phase unwrapping algorithm. Radial basis functions (RBFs) were fit to the raw Eulerian displacements and analytical spatial derivatives were computed directly from the coefficients of the RBFs. Using these derivatives, a 3D deformation gradient tensor and subsequently a 3D Cartesian Lagrangian strain tensor could be computed at any point within the myocardium.

The geometry of both the RV and LV was defined by fitting a triangular surface mesh to the manually-delineated endocardial contours at end-diastole (Figure 1B). The local coordinate system was defined for any point on the mesh: the radial direction was the inward normal of the endocardial mesh, the longitudinal direction was tangent to the surface and pointed towards the manually-defined ventricular apex, and the circumferential direction was the cross product of the radial and longitudinal directions. The Cartesian strain tensors were transformed to correspond with the local coordinate system to obtain radial, circumferential, and longitudinal strains (Err, Ecc, and Ell). Torsion was defined as the circumferential-longitudinal shear angle. Regional activation times were computed by performing cross-correlation between regional 2^nd principal strain curves and the average curve.

Ecc varied regionally within the RV with the lowest values (16%) in the outflow region (Figure 2A). Ell varied considerably around the circumference of the RV (14 – 22%) with global Ell being higher in the RV relative to the LV (Figure 2B). RV torsion was found predominantly in the lateral region and was comparable to LV torsion (Figure 2C). Regional activation times indicated that the RV as a whole contracted later than the LV with the lateral wall of the RV contracting last (Figure 2D). 3D Spiral cine DENSE combined with local coordinate system-based post-processing is capable of resolving the complex geometry and 3D motion of both the right and left ventricles. In healthy subjects, regional variations in Ecc and Ell exist within the RV, however they have comparable global magnitudes. RV torsion was similar to torsion seen in the LV and the RV was found to contract later than the LV. Although this study provides insight into normal biventricular function, future studies should seek to understand how this function is disrupted by disease.