Three-dimensional regional strain analysis in porcine myocardial infarction: a 3T magnetic resonance tagging study

Sahar Soleimanifard, Khaled Z. Abd‐Elmoniem, Tetsuo Sasano, Harsh Agarwal, M. Roselle Abraham, Theodore P. Abraham, Jerry L. PrinceJournal of Cardiovascular Magnetic Resonance 2012 · DOI

Abstract

Cumulative 3D strain information accurately identifies infarcts and their neighboring regions from healthy myocardium. The 3D interrogation of myocardial contractility provides incremental diagnostic accuracy in delineating the dysfunctional and nonviable myocardium in comparison with 1D or 2D quantification of strain. The infarct neighboring regions are the major beneficiaries of the 3D assessment of regional strain.

Introduction

When a heart attack (myocardial infarction, or MI) occurs, the affected muscle tissue stops contracting normally. Clinicians use Cardiovascular Magnetic Resonance (CMR) to measure "strain"—the deformation of the heart muscle—to assess damage and viability. However, most current clinical practices rely on 1D (circumferential) or 2D (circumferential and radial) strain measurements. Because the heart is a complex 3D structure that twists, shortens, and thickens simultaneously, 2D slices often miss out-of-plane motion. This is particularly problematic in the "adjacent" or border zones—areas next to the dead scar tissue that might still be salvageable. This study demonstrates that a full 3D interrogation of the heart provides significant incremental value over traditional 1D and 2D models.

Methods and Datasets

The researchers utilized a porcine model, surgically inducing MI in 20 farm pigs (15 survived the full protocol) to simulate human heart disease. Imaging was performed at baseline, 11 days post-MI (early), and 1 month post-MI (late) using a 3.0 Tesla MR scanner.

The core innovation was the use of z-encoding harmonic phase (zHARP) imaging. This technique allows for the tracking of 3D motion from a single image plane by encoding "tags" into the tissue that move as the heart beats. The heart was divided into segments based on the transmurality of the scar as seen on Late Gadolinium Enhancement (LGE) imaging, creating a labeled dataset of "Infarct," "Adjacent," and "Remote" (healthy) tissue.

Figure 5: (A) Segmentation of the myocardium into Infarct, Adjacent, and Remote zones. (B) The zHARP tagging sequence capturing myocardial motion from 130ms to 930ms.

The researchers calculated both directional strains (circumferential, longitudinal, and radial) and principal strains ($E_1, E_2, E_3$). Principal strains represent the primary directions of maximum and minimum deformation and are independent of the imaging plane. Univariate (1D), bivariate (2D), and multivariate (3D) logistic regression models were then built to distinguish between the tissue types.

Figure 6: Representative MR images from different subjects showing the location of the myocardial infarction (yellow arrows).

Key Insights

  • The "Adjacent Zone" Advantage: The most striking finding was that 3D strain is most useful in the "gray zone" surrounding the infarct. While dead tissue is relatively easy to spot even with 1D tools, the subtle dysfunction in neighboring regions requires the full 3D picture to be accurately identified.
  • Principal Strains as Universal Metrics: Principal strains proved highly effective because they represent the actual physics of muscle deformation regardless of the viewing angle or slice alignment.
  • Longitudinal Motion Matters: By moving from 2D to 3D, the researchers could account for longitudinal shortening. Ignoring this third dimension leads to a significant "blind spot" in diagnostic accuracy.

Results

Receiver Operating Characteristic (ROC) analysis compared the diagnostic accuracy of the models:

  • Adjacent Segment Identification: The 3D model achieved 96% accuracy (c-statistic: 0.981), significantly outperforming the 2D model (81% accuracy, c-statistic: 0.941) and the 1D model (71% accuracy, c-statistic: 0.846).
  • Infarct Identification: All models performed well, but the 3D model still led with 98% accuracy compared to 95% for 1D.
  • Early Detection: While wall thickening showed no significant improvement in adjacent segments early on, 3D strain detected functional changes immediately.

Conclusion and Future Work

This study highlights that 3D regional strain analysis significantly outperforms traditional methods in identifying damaged heart tissue, particularly in the critical border zones. While the porcine model primarily produced transmural infarcts, the benefits of 3D strain might be even more critical for sub-endocardial infarcts.

Future work should focus on clinical translation. While 3T MRI is common, the zHARP sequence and complex post-processing are not yet standard. Eventually, 3D strain could be used to predict patient response to salvage therapies or monitor the success of regenerative treatments like stem cell therapy.

Summary updated

Abstract

Cumulative 3D strain information accurately identifies infarcts and their neighboring regions from healthy myocardium. The 3D interrogation of myocardial contractility provides incremental diagnostic accuracy in delineating the dysfunctional and nonviable myocardium in comparison with 1D or 2D quantification of strain. The infarct neighboring regions are the major beneficiaries of the 3D assessment of regional strain.
BibTeX
@article{soleimanifard2012three,
  author = {Sahar Soleimanifard and Khaled Z. Abd‐Elmoniem and Tetsuo Sasano and Harsh Agarwal and M. Roselle Abraham and Theodore P. Abraham and Jerry L. Prince},
  title = {{Three-dimensional regional strain analysis in porcine myocardial infarction: a 3T magnetic resonance tagging study}},
  journal = {Journal of Cardiovascular Magnetic Resonance},
  year = {2012},
  doi = {10.1186/1532-429x-14-85},
}
Figure