Oral Abstract 2: In Vivo Quantification of the Mean Microstructural Deformation of the Myocardium using DENSE MRI

Presenting Author(s)

• TH

  Tawfik Hussein, MSc

  Ph.D. Candidate
  Emory University & Georgia Institute of Technology

Primary Author(s)

• TH

  Tawfik Hussein, MSc

  Ph.D. Candidate
  Emory University & Georgia Institute of Technology

Co-Author(s)

• JO

  John Oshinski, PhD

  Professor of Radiology and Biomedical Engineering
  Emory University

Background: Capturing intramyocardial deformation is central to understanding its biomechanical behavior in both health and disease as it can elucidate local changes that global indices fail to quantify. While deformation is usually quantified by measuring myocardial strain, most current strain measures fail to capture changes in the myocardial microstructure, reducing their clinical utility. Because displacement encoding with simulated echoes (DENSE) MRI can measure pixel-wise displacement, it carries the potential to reveal local microstructural deformations as well as enable registration with diffusion tensor imaging (DTI). The goal of this work was to use DENSE to locally quantify the mean volume strain, fiber shortening, laminar thickening, in-plane laminar alignment, laminar sliding, and out-of-plane laminar alignment strains (α₁, α₂, α₃, α₄, α₅ and α₆ respectively) in vivo, which relate directly to changes in the myocardial microstructure. [1]. We hypothesized that DENSE could capture the mean microstructural deformations within the myocardium, and laminar thickening strain would be greater than the fiber shortening strain.

Methods: In 10 healthy humans (6 males ages 25-59 years and 4 females ages 22-31 years), ECG-gated cine DENSE was used to measure displacement from the short-axis (SA) and long-axis (LA) views of the LV myocardium. The displacements from both views were combined to calculate α₁, α₂, α₃, α₄, α₅ and α₆ to capture the full 3D kinematics in the SA slices, as seen in Figure 1 [1, 2]. Subsequently, the SA images were divided based on the AHA 16-segment model. For each segment, the measured strains were averaged, and their respective mean and standard deviation were calculated. Finally, the peak systolic strain was calculated in all 3 myocardial SA regions and plotted segmentally in an AHA bullseye plot.

Results: Shown in Figures 2 & 3, α₂ and α₃ revealed that the myocardium undergoes greater laminar thickening than fiber shortening, with the lateral wall having more deformation than the septal wall. Additionally, out of the 3 shear strains, α₆ was found to have the greatest contribution. Indeed, this is consistent with the current understanding that the myocardial laminae undergo significant thickening compared to fiber shortening, a process that is mainly driven by the transverse alignment of the laminae, captured by α₆ [3]. Interestingly, α₁ showed that the myocardium decreases in volume during systole, which challenges the incompressibility constraint that is commonly assumed for the myocardium.

Conclusion: Capturing 3D deformation along the myocardial material directions allows quantification of microstructural changes, enabling intuitive characterizations of cardiac disease. These strains could provide clinicians with more mechanistic insights into the LV remodeling process and serve as predictive markers of adverse remodeling in various diseases. Additionally, because these DENSE-derived strain attributes capture the voxel-wise deformation of the microstructure, they are better suited for registration with DTI.

Figure 1. Computational workflow for calculating the strains from DENSE images. Each volunteer underwent a DENSE where 3 SA and 3 LA views were obtained. They were then divided into segments based on the AHA 17-segment model to register the through-plane displacements from LA views with the SA in-plane displacements. This obtains 3D displacement for each segment, which were then used to calculate the strains.

Figure 2. (A) The mean α1 (volume strain) and standard deviation (n = 10) for the basal SA view. (B) The mean α2 (fiber strain) and standard deviation for the basal SA view. (C) The mean α3 (laminar thickening strain) and standard deviation for the basal SA view. (D) The mean α4 (in-plane alignment strain) and standard deviation for the basal SA view. (E) The mean α5 (laminar sliding strain) and standard deviation for the basal SA view. (F) The mean α6 (out-of-plane alignment strain) and standard deviation for the basal SA view.

Figure 3. AHA bullseyes for (A) The mean α1 (volume strain) in the basal SA view. (B) The mean α2 (fiber strain) in the basal SA view. (C) The mean α3 (laminar thickening strain) in the basal SA view. (D) The mean α4 (in-plane alignment strain) in the basal SA view. (E) The mean α5 (laminar sliding strain) in the basal SA view. (F) The mean α6 (out-of-plane alignment strain) in the basal SA view.