Introduction

Cardiovascular disease (CVD) remains a leading cause of death worldwide accounting for 33% or 1 in every 3 deaths in the United States since 2008 [1]. The American Heart Association plans to reduce deaths caused by CVD by 20% and improve the cardiovascular health of all Americans. In order to achieve these goals, the ability to diagnose, treat, and manage current patients and individuals at risk must be improved. To date, our understanding of cardiac function and pathology is still limited, and various techniques have been developed for the effective diagnosis, treatment, and management of CVD. MRI has been an active platform for the development of several non-invasive methods used to understand myocardial structure and function. Heart disease can affect any region of the myocardium and the complex movement and contraction of the heart; various CMR imaging techniques have been developed to provide qualitative and quantitative analysis of contractile function. Technical challenges for imaging myocardial strain include rapid data acquisition, rapid image analysis, accuracy, and reproducibility. Imaging myocardial strain is of growing importance in the assessment of CVD. Echocardiographic studies have shown myocardial strain is more predictive of future, adverse cardiac events than left-ventricular (LV) ejection fraction (EF) for patients with known or suspected LV impairment [2, 3]. Such findings and advancements in echocardiography are influencing cardiac magnetic resonance (CMR) imaging where interest in myocardial strain imaging is increasing. Various studies assessing myocardial strain have shown that MR strain imaging has the potential to demonstrate subclinical myocardial changes prior to the onset of myocardial dysfunction. Examples of such studies include impaired contractile function in obese children [4] and systolic dysfunction in patients with Type 2 diabetes Mellitus [5]. Furthermore, CMR has the potential to predict changes in LV outcomes as shown by Mangion et al. in patients who have suffered acute myocardial infarction [6] and by Auger et al. in heart failure (HF) patients undergoing cardiac resynchronization therapy (CRT) [7]. Various CMR strain imaging techniques are available which include myocardial tagging, harmonic phase MRI (HARP), phase contrast velocity encoding (PC-MR), and displacement encoding with stimulated echoes (DENSE). Understanding the difference between the advantages and limitations of each is important.

Strain and strain imaging methods

Strain is a geometrical measure of deformation representing the relative displacement between particles in a material body; i.e. a measure of how much a given displacement differs locally from a rigid-body displacement. The myocardial strain tensor is described by three principal strains which can be assessed by the direction of deformation, namely, radial thickening (Err), circumferential shortening (Ecc) and longitudinal shortening (Ell). Figure 1(A) illustrates the 3-dimensional deformation strains in the LV, while Figure 2(B, C) illustrates the example of 2-dimensional deformation. Myocardial tagging is an established technique has been considered to be the gold standard of strain imaging. Tagging allows for the spatial distribution of saturated magnetization across the myocardium producing dark bands which are also known as tag lines [8]. As the heart contracts, these tag lines become distorted reflecting the underlying tissue motion and thus providing features which allow for temporal tissue tracking methods. Tracking methods allow for displacement and strain calculations in regional segments of the LV. Figure 2 illustrates the distortion of the tag lines from time of encoding to a systolic time point. The corresponding Ecc and Err strain maps are included for a healthy volunteer. Myocardial tagging is an established technique used clinically in the detection of viable myocardium, cardiac dyssynchrony, and ischemia. However, quantitative results are time consuming and requires substantial user interaction hindering its use in a day to day clinical setting. Furthermore, the spatial resolution is determined by the distance between tag lines and not the images. Harmonic phase (HARP) MRI is an imaging technique used for the rapid analysis and visualisation of tagged MR images [9]. The method is based on the fact that spatial modulation of magnetization (SPAMM)-tagged MR images consist of a collection of distinct spectral peaks in the Fourier domain where each peak contains motion information in a certain direction. As each peak is isolated, the inverse Fourier transform is calculated, resulting in a phase linearly related to the tissue motion in that direction. HARP processing can be up to 10 times faster than conventional tag following techniques, however, isolating spectral peaks in k-space leads to SNR loss na decreased spatial resolution. PC-MR is an imaging technique which encodes the instantaneous tissue velocity directly into the voxel phase with the use of a bipolar gradient [10]. PC-MR has the potential to provide valuable information in the evaluation of global and regional systolic and diastolic function. The sequence is available on most clinical scanners and has fast post-processing methods. Velocity and strain rate are determined from phase velocity maps. PC-MR has been validated and is often used as a technique for measuring blood flow. However, phase velocity mapping has been used in several clinical studies to quantify regional myocardial motion and detect contractile dysfunction under various conditions. Time to peak velocity is used to quantify ventricular dyssynchrony and reduced systolic and diastolic peak velocities to quantify HF. Strain can be calculated by integrating velocities with respect to time, however, as errors propagate through the integration, this is not often done. Strain analysis of cine steady-state free precession (SSFP) otherwise known as feature-tracking (FT) offers the advantage of convenience, since strain is assessed from standard SSFP images which are part of nearly all standard cardiac MR examinations, thus there is no need for additional acquisitions. FT involves tracking features in the SSFP images such as the endocardial border through the cardiac cycle from which displacement fields are derived and strain is calculated. There are various algorithms such as Tom Tec and heart deformation analysis (HDA) have been implemented to calculate strain from SSFP images. These methods are fast and automatic and have been validated against manual delineation by expert observers [11] and have been evaluated in healthy volunteers [12]. Results show that whole slice global strain measured using feature-tracking have good correlation to myocardial tagging derived global strain. However, these methods have not been evaluated in patient populations and various studies have disregarded regional strain assessment using SSFP FT [13, 14] due to large regional differences [12] and poor reproducibility [15, 16] Cine DENSE MR is becoming the gold standard for strain imaging techniques. Cine DENSE encodes the tissue displacement directly into the phase of the images, therefore, myocardial motion and deformation is being measured directly at a high spatial resolution over segments of the cardiac cycle [17, 18]. DENSE MR imaging contains a SPAMM kernel to position encode the magnetization at an end diastolic time frame. Tissue displacement which occurs between displacement encoding pulses and subsequent acquisition times results in a phase shift of the stimulated echo. Proceeding post-processing methods, this phase shift is a linear representation of tissue displacement. Tissue tracking methods are applied to calculate LV displacement fields and strain. Figure 3 illustrates the magnitude, phase in two encoding directions and the corresponding vector displacement field at end-systole. The Ecc strain time curves and DENSE Ecc strain map in Figure 4 A and B, illustrates the contraction pattern of a healthy volunteer. However Figure 4 D and E, decreased contractile function is clear in both strain time curves and the Ecc DENSE strain maps in the inferior-septal region. This corresponds well with the late gadolinium enhanced image which clearly shows myocardial scar in this region. Figure 5 illustrates an example where cine DENSE strain is used to calculate and quantify regions of delayed mechanical activation in HF patients undergoing CRT. The strain time curve in Figure 5(A) indicates delayed activation in the lateral wall of the LV. Active contour methods applied to a strain matrix detects regions of early and delayed mechanical activation times along the LV as shown in Figure 4(B). Bull’s-eye plot of multi-slice activation mapping are produced and can be used to guide the CRT procedure to ensure the LV lead is placed in a region of maximal delayed mechanical activation.

Conclusions

The importance of strain imaging is growing significantly and studies are demonstrating the ability of strain to diagnose and manage a wide range of cardiovascular disease. There are various methods available, each with its advantages and limitations. Each study should choose the methods carefully depending on speed, accuracy and dependency.

Acknowledgements

No acknowledgement found.

References

1. Roger, et al., Cir, 2012; 125: e2-e220 2. Mignot, et al. J Am Soc Echocardiogr., 2010. 23(10): p. 1019-1024. 3. Stanton, et al. Circ: Cardiovasc Imaging, 2009. 2(5): p. 356-364. 4. Jing et al. J Cardiovasc Magn Reson, 2016; 18:28 5. Ernande et al. Rdiology, 2012; 265:402-409 6. Mangion, et al.. 2017, J Cardiovasc Magn Reson, conference proceedings. In print. 7. Auger et al. J Mang Reson, 2017. 8. Axel, et al.. Medical Image Analysis, 9(4):376_393, 2005. 9. Osman, J of the Soc of Magn Reson in Med, 42(6):1048, 1999 10. Pelc, et al. J of Magn Reson Imaging, 5(3):339_345, 1995. 11. Li et al. JACC, 2010; 3(8):860-6 12. Augustine et al. JCMR, 2013;15:8 13. Moody et al. JMRI, 2015; 41:1000–1012 14. Hor et al. JACC Img, 2010; 3(2):144-151 15. Morton et al. JCMR, 2012;14:43 16. Wu et al. J Cardiovasc Magn Reson. 2014; 16(1): 10). 17. Aletras, et al. J of Magn Reson, 137(1):247_252, 1999. 18. Kim et al. Radiology, 2004; 230:826-871