Translating High-Frame-Rate Imaging into Clinical Practice: Where Do We Stand?

High-frame-rate imaging enables a new method for estimating the elastic properties of the myocardium (i.e., stiffness): shear wave imaging (SWI)^[18–21] can measure and visualize certain vibrations in the myocardium (shear waves, SW) that propagate at a velocity that is directly related to its stiffness. In a large, isotropic, elastic medium, the velocity of shear waves (c) directly relates to the shear modulus of the medium (i.e., its stiffness, µ), and its density (ρ).^[22]

Although these assumptions do not completely hold for myocardium (which is neither large nor isotropic), this approach allows for the first time a noninvasive, destruction-free, direct estimation of tissue stiffness in vivo. SWI has already been applied in various fields of medicine, including breast lesions detection,^[23] liver fibrosis staging,^[24] and assessment of arterial stiffness,^[25] and is commercially available as an additional application to noncardiac ultrasound devices. Cardiac SWI is technically much more challenging as the result of a smaller transducer footprint, intercostal sector imaging, and the continuous motion of the heart. Technical advances have only recently allowed the translation of this principle to cardiac applications. Initially, studies investigating the use of SWI to quantify myocardial stiffness were mainly performed in open-chest animal models.^[28–30] Only recently, has this method become feasible in closed-chest, in vivo animals and humans.^{[28,29,31,32]}

Natural shear waves: SWs are naturally induced in the heart by physiologic events, for example, by the closure of the aortic or mitral valve (Figure 2). Natural SWs cause a relatively large tissue displacement, which can be detected with transthoracic high-frame-rate imaging so that the approach is suitable for closed-chest myocardial stiffness estimations. The main disadvantage of this method is that natural SWs only occur immediately after valve closures, which limits stiffness measurements to only two time points in the cardiac cycle, that is, early isovolumetric contraction and relaxation.^[33]

Figure 2

So far, researchers have mainly investigated natural SWs in parasternal, long axis images of the interventricular septum because of the advantageous anatomical position of this cardiac structure.^[13,34] Note that 2D imaging of SWs measures only the projection of the SW velocity vector on the image plane. A recent study on three healthy volunteers (HV) demonstrated the feasibility of estimating myocardial SW velocities in 3D along different directions.^[35] An example of a mechanical wave in the heart visualized with fast 3D imaging technology is given in Figure 3.^[36]

Figure 3

Artificially induced shear waves: SWs can also be generated by applying a high-intensity, focused ultrasound beam (“push pulse”) ^[37] (Figure 4). An advantage of this method is that the shear wave can be induced at a chosen position and any time point during the cardiac cycle^[30] so that stiffness can be estimated where and when it is of physiologic interest and free from potential interference from changing loading conditions or tissue activation. However, push pulses require a high power output from the transducer in order to produce tissue displacements, thus raising safety concerns. Furthermore, the excited SWs are small in amplitude (∼1 μm^[30] vs. 100 µm^[28] in natural SWs) and attenuate quickly when propagating through the tissue, which makes their detection challenging and allows stiffness estimates only under ideal scanning conditions. Consequently, studies utilizing induced shear waves have so far mostly been performed in children or open-chest animals).^[30,36,38]

Figure 4

Normal values: Several studies have shown the propagation speed of natural SWs to be in the range of 3–5 m/s in healthy volunteers^[33,34,39] with a lifelong increase of SW velocities due to physiological alterations in the cellular and extracellular matrix that impact myocardial stiffness with aging.^[40] A recent study by the Leuven research group that measured SWs originating naturally after aortic valve closure (AVC) and mitral valve closure (MVC), recruited 104 healthy patients (age range 2–18 years old) and reported SW velocities of 3.0 ± 0.3 m/s after AVC and 2.8 ± 0.3 m/s after MVC. Petrescu et al. found a mean velocity of 3.75 ± 0.78 m/s for AVC vs. 3.07 ± 0.51 m/s for MVC in the 20- to 39-year-old group, with a significant increase up to 3.75 ± 0.78 m/s vs. 3.84 ± 0.79 m/s in the 40- to 59-year-old-group, and 4.50 ± 1.13 m/s vs. 4.33 ± 0.61 m/s in the 60- to 80-year-old group (p < 0.01 between groups), respectively.^[31] Similarly, Villemain et al. showed a gradual increase in myocardial stiffness calculated using SW velocities produced by acoustic radiation force (2.59 ± 0.58 kPa for the 20- to 39-year-old group, 4.70 ± 0.88 kPa for the 40- to 59-year-old HV group, and 6.08 ± 1.06 kPa for the 60- to 79-year-old HV group, p < 0.01).^[4,7]

Assessment of diastolic and systolic properties of the myocardium: SWI enables the measurement of the intrinsic properties of myocardium, reflecting its stiffness and contractile state.^[31,41] During mid- and late diastole, SWI would therefore measure pure elastic properties. At all other time points, contractile forces add up to a varying extent.^[40]

Repeated application of ultrasonic push pulses in Langendorff-perfused, isolated rats with generation of artificial SWs through the entire cardiac cycle, showed dynamic stiffness variations during systole and diastole (Figure 4). Moreover, myocardial stiffness measured during systole was found to be strongly dependent on the inotropic state modulated with calcium or beta-adrenergic agonists, indicating that peak systolic myocardial stiffness has the potential to be used as an index of myocardial contractility.^[30] The same author showed that end-diastolic stiffness measured by SWI using artificial SWs could also be used to determine the passive diastolic properties of the myocardium in an open-chest animal study.^[43] Myocardial stiffness was also quantified in explanted porcine heart grafts (n = 14) to predict viability.^[44] Good correlations could be found between SW velocities and cardiac function parameters (left ventricular pressure, heart rate, myocardial contractility and relaxation rate, and end diastolic pressure) suggesting that SW velocity could be used as a marker of graft viability. In an in vivo animal experiment, Bézy et al. found that SW velocity after MVC was influenced by both acute changes in loading and myocardial mechanical properties, and was strongly related to chamber stiffness, invasively derived by pressure-volume loop analysis.^[45] A clinical study using natural SWs reported a positive correlation between the SW velocity after AVC and end-systolic elastance after bicycle exercise and dobutamine administration.^[46] Another clinical study showed that SWs after MVC have a strong correlation with invasively measured end-diastolic filling pressures and can better predict elevated end-diastolic filling pressures than the current guideline-recommended algorithm.^[47] It should be recalled, however, that the assessment of natural SW velocities is limited to the time point of the physiologic event, i.e., the closure of the aortic or mitral valve, which always leaves an uncertainty about the contractile state in which the heart is at the time point of the measurement.

Local variations of myocardial stiffness: Various studies have shown that SWI may be able to assess structural inhomogeneities that result in local variations of their propagation velocity.^[43] As such, SWI could potentially allow the early diagnosis of conditions associated with localized variations in tissue stiffness inside the myocardium such a local ischemia, inflammation, or infiltration that could be otherwise missed because their diagnosis is challenging with standard echocardiography.

Pernot et al. evaluated the potential of SWI using acoustic radiation force to quantify passive diastolic myocardial stiffness in an ovine model of ischemic cardiomyopathy and showed that SWI can differentiate between a stiff infarcted wall and a softer wall containing stunned myocardium.^[43] Wouters et al. reported that the distinction of a scarred, nonviable septum and a thin, dysfunctional, remodeled septum in heart failure patients with left bundle branch block (LBBB) is possible using natural SWI after MVC.^[48]

Hypertrophic cardiomyopathy (HCM) is a genetic disorder characterized by ventricular hypertrophy associated with other histopathological alterations such as myocyte disarray, small vessel disease, and interstitial as well as replacement fibrosis.^[49] Villemain et al. investigated the diastolic myocardial stiffness in 20 HCM patients with heart failure and preserved ejection fraction (HFpEF) and 60 healthy volunteers of three age groups (20 to 39 years, n = 20; 40 to 59 years, n = 20; and 60 to 80 years, n = 20) in whom SWs were induced using acoustic radiation force at end-diastole.^[22] They found that end-diastolic myocardial stiffness was significantly increased in HCM patients (12.7 ± 2.9 kPa vs. 4.5 ± 1.7 kPa; p < 0.01) compared to healthy volunteers. The authors reported the optimal cut-off value of myocardial stiffness for detection of HCM-HFpEF to be 8 kPa (Figure 5). In addition, receiving operator characteristic (ROC) analysis revealed SWI has an excellent ability to distinguish HCM patients from healthy volunteers with an area under the curve of 0.99. In addition, myocardial stiffness increased significantly with age (2.59 ± 0.58 kPa, 4.70 ± 0.88 kPa, and 6.08 ± 1.06 kPa, respectively; p < 0.01 between each group) and showed a significant correlation with classic echocardiographic parameters of diastolic function (E/A ratio, r = 0.55; left atrial volume index, r = 0.67; E/e’ ratio, r = 0.62; E/Vp ratio, r = 0.50). Similar findings were shown by Strachinaru et al. using a clinical color TDI application with a frame rate of over 500 Hz. Both MVC- and AVC-induced SWs had an increased velocity in patients with HCM (n = 20) compared to gender- and age-matched healthy volunteers (MVC: 6.9 ± 1.1 m/s vs. 4.7 ± 0.8 m/s; p < 0.001, AVC: 5.1 ± 0.7 vs. 3.6 ± 0.5 m/s; p < 0.001).^[50] SW velocity after AVC also showed excellent diagnostic accuracy in patients with HCM (AVC = 0.99). Furthermore, increased end-diastolic stiffness assessed with SWI was related to increased native T1 time, a marker of fibrosis (r = 0.71; p < 0.01). In addition, tissue architecture in HCM patients was evaluated by analyzing fractional anisotropy (FA) using SWI. A significant decrease in FA was found in HCM patients compared to healthy volunteers, reflecting changes in tissue architecture in these patients. The same group correlated the velocity of natural SWs with the presence of scarred tissue in 10 HCM patients with prior septal ablation therapy, as a model for local increase of stiffness. A difference between various degrees of pathological stiffness (nonablated HCM myocardium vs. myocardial scar tissue in ablated segments) could be detected.^[51]

Figure 5

Cardiac amyloidosis is a group of infiltrative disorders characterized by extracellular deposition of amyloid fibrils in organs and tissues.^[52] The prognosis is poor, especially when there is cardiac involvement. A study by Petrescu et al. has shown that natural SW velocity after MVC and AVC was significantly increased in patients with cardiac amyloidosis (n = 18) in comparison to healthy volunteers (n = 63) (MVC: 6.3 ± 1.6 m/s vs. 3.6 ± 1.0 m/s; p < 0.001, AVC: 5.6 ± 1.1 m/s vs. 3.8 ± 0.7 m/s; p < 0.001).^[31] Moreover, SW velocity was related to the grade of diastolic dysfunction in these patients and had a strong correlation to E/e’ ratio (r = 0.74; p < 0.01), a marker of left ventricular filling pressures.

In hypertensive heart disease (HHD), the left ventricle (LV) becomes hypertrophied in response to chronic arterial hypertension (concentric remodeling).^[53] At a later stage of disease, structural changes dominate, resulting in the accumulation of collagen within the interstitium (concentric hypertrophy). A study by Cvijic et al. showed that SW velocity after MVC was significantly increased in patients with HHD (n = 33) compared to age-matched healthy volunteers (n = 26) (5.83 ± 1.2 m/s vs. 4.04 ± 0.96 m/s; p < 0.001).^[54] Furthermore, SW velocity was shown to vary depending on the stage of disease. Patients with concentric hypertrophy (excessive collaged accumulation) showed significantly elevated SW velocity (6.3 ± 1.1 m/s; p < 0.001), while patients with concentric remodeling (normal LV mass, abnormal relative wall thickness) had values similar to the healthy volunteers (4.8 ± 0.7 m/s; p = 0.075). These differences were remained present, even after accounting for potential variations in wall stress between patient groups. An optimal cut-off value of ≥ 5 m/s could detect HHD patients with concentric hypertrophy with a sensitivity of 94% and specificity of 91%. These findings are in line with previous studies reporting that interstitial fibrosis is only present in patients with concentric hypertrophy, which suggests that SWI could reveal structural and functional differences among patients with HHT.

A recent study also suggested a potential role for SWI in diagnosing graft dysfunction, one of the most common complications after heart transplant (HTx).^[55] Histological changes accompanying allograft dysfunction include myocyte hypertrophy and diffuse myocardial fibrosis, which can in turn lead to diastolic dysfunction due to increased myocardial stiffness.^[56] Clinical evaluation of graft dysfunction therefore aims at tissue characterization and assessment of diastolic function. Given that allograft dysfunction may lead to a stiffer myocardium, SWI might aid in the early detection of this condition. A study by Petrescu et al. showed that in 52 HTx recipients SW velocity at MVC correlated better with native myocardial T1 values (r = 0.75; p = 0.0001) than invasive measurements of pulmonary capillary wedge pressure (r = 0.54; p < 0.001).^[32] These findings suggest that cardiac SWI has the potential to become a fast and cheap alternative to magnetic resonance imaging and invasive procedures in the follow-up of HTx patients (Figure 6).

Figure 6

The quantitative estimation of myocardial stiffness by SWI without the need for invasive procedures, cardiac magnetic resonance, or histology has been shown to be feasible for characterizing diffuse myocardial fibrosis, ischemic scar, and other pathologic conditions in addition to the noninvasive assessment of myocardial systolic and diastolic function. Further clinical validation is needed, however, before this promising technique can be applied in the clinical routine. Nevertheless, current preliminary results are promising and suggest that SWI will complement our echocardiographic toolbox in the coming years.

Global and segmental measurements of myocardial strain provide additional information on cardiac mechanics beyond conventional parameters such as ejection fraction and visual wall motion analysis. Strain can be obtained by tissue Doppler or speckle tracking echocardiography (STE). ^[39] While STE is more feasible in the clinical routine than tissue Doppler, it suffers from the low frame rate of the underlying grey scale images (40–80 fps). This results in under-sampling of fast and short-lived events, e.g., in the isovolumic phases of the cardiac cycle and the in accuracy of temporal derivatives of motion and strain, i.e., velocity and strain rate.^[57,58] Recently, dedicated high-frame-rate STE algorithms have been developed in order to combine the high temporal resolution of ultrafast ultrasound imaging with the comfort of 2D speckle tracking algorithms.^[57,59–61] Joos et al. applied STE on high-frame-rate B-mode images at 500 fps that were obtained by special coherent compounding methods based on motion compensation. A global longitudinal strain of the left ventricle was obtained from STE in 10 subjects and compared to the results provided by a clinical scanner. Interestingly, group means were not statistically different (p-value = 0.33).^[59] Orlowska et al. proposed a two-step 2D STE algorithm based on cross-correlations purposely developed for high-frame-rate imaging echocardiography. The method was first optimized and validated on simulated data and then tested in vivo on ten healthy volunteers, thus demonstrating its clinical applicability and feasibility. The estimated peak global longitudinal strain values were compared with those measured with tissue Doppler data from a clinical scanner showing good correlation and negligible differences (-20.94% vs. -20.31%, p-value = 0.44), and were in the same range with values reported in the literature.^[61]

These results show that dedicated high-frame-rate STE algorithms allow speckle tracking at frame rates that so far only tissue Doppler could provide. Studies need to show how clinical decision-making can benefit from such temporally resolved myocardial deformation information.^[2,59,60]

Noninvasive imaging of the coronary vasculature using ultrafast ultrasound: Over the past few years, significant progress has been made in noninvasive imaging of the coronary arteries with computed cardiac magnetic resonance imaging^[62] or coronary computed tomography angiography.^[63] These modalities are limited, however, to the exploration of the anatomy of large coronary arteries. Small-vessel function can only be indirectly assessed by measuring flow changes in the larger arteries.^[64,65]

Coronary ultrafast Doppler angiography (CUDA) was developed as a specific application for imaging coronary arteries enabling the noninvasive imaging of the coronary vasculature and the assessment of coronary flow reserve. The technique involves acquiring ultrasound images of the beating heart at high frame rates and using specific spatiotemporal clutter filters^[66] in order to distinguish the myocardium with high echogenicity and relatively slow motion from blood vessels with low echogenicity and faster motion.^[67] Thus, CUDA is able to visualize the coronary vasculature down to 100 µm with high sensitivity and without the use of contrast agents. It can also quantify intramyocardial blood flow changes in a wide range of flow conditions (1 mm/s to 1 m/s). The first experiments in the beating heart of large animal models that showed that CUDA accurately quantifies the change of flow rate in coronary arterioles (with diameters ranging from 500 to 100 µm) during hyperemia were performed by Maresca et al.^[66] The authors compared the measurements to the gold standard by placing a proximal flow meter probe on the epicardial artery. Quantification of the coronary flow changes during hyperemia was in good agreement with gold standard measurements (r² = 0.89), as well as the assessment of coronary flow reserve (2.35 ± 0.65 vs. 2.28 ± 0.84; p = NS). In the infarcted animals, CUDA images revealed the presence of strong hyperemia and the appearance of abnormal coronary vessel structures in the reperfused LAD territory. Figure 7 shows long-axis and short-axis images of the coronary vasculature superimposed on B-mode image of the myocardium. The 2D Doppler approach, however, does not allow a reliable quantification of the coronary flow rate due to the angle dependence of Doppler estimates. To solve this limitation, 3D ultrafast CUDA was recently introduced to image the coronary vasculature in 3D and quantify the absolute coronary volumetric flow rates noninvasively.^[68] The feasibility of the technique was demonstrated in an open-chest swine model. Figure 8 represents the coronary blood flow at baseline and during reactive hyperemia after transient occlusion of the left anterior descending (LAD) artery.

Figure 7

Figure 8

CUDA provides the detection of coronary flow in epicardial and intramural arteries with characterization of coronary flow reserve and intramural perfusion without the use of contrast agents or exposure to ionizing radiation. This technique could be used, for example, in patients with microvascular angina, but it will need further clinical validation and study.

Intracardiac flow visualization: Color flow imaging is the conventional method to estimate velocities displayed together with the B-mode image. However, it can only represent the flow component in the direction of the ultrasound beam and is limited by aliasing effects.^[69] Speckle tracking flow has been attempted in regular echocardiographic 2D B-mode imaging with the use of contrast agents. This is called “echocardiographic particle image velocimetry” (echo-PIV), and it can assess blood motion in any direction within the B-mode image plane.^[70,71] The technique, however, suffers from the varying and limited density of contrast reflectors and the too low frame rate of the regular grey scale imaging, which hardly allows the resolution of flow velocities beyond 40 cm/s.^[64]

The recent development of high-frame-rate ultrasound made a new approach to blood flow assessment possible. High-frame-rate imaging-based blood speckle tracking (BST) is a novel blood flow visualization that relies on tracking native blood on high-frame-rate images without the need for contrast agent injections. It can be used for the evaluation of blood flow energetics by describing the blood vortex inside the ventricles^[65,72,73] or to visualize and quantify flow in congenital defects, valve disease, or pulmonary hypertension (Figure 9). Vortex visualization in fetal cardiac ventricles could also contribute to a better evaluation of cardiac function before birth and might improve prenatal detection of coarctation of the aorta. Furthermore, BST allows the visualization of altered flow patterns in ventricles with reduced function and by using high-frame-rate flow data in the calculation of intraventricular pressure gradients.^[74]

Figure 9

BST is so far the first commercially available application of high-frame-rate imaging in cardiology and has mainly been used to image cardiac blood flow patterns in neonates^[72] and pediatric patients.^[65] The feasibility of this method was also investigated in the adult heart (Figure 10),^{[36,70,75,76]} but image depth is limited to approximately 8 cm (Nyrnes). An increase in depth would require a probe with a lower frequency, leading to reduced spatial resolution and, thus, providing insufficient tracking results.

Figure 10

Intracardiac blood flow analysis may contribute to a deeper understanding of complex flow physiology. However, further studies are needed to evaluate the potential of this novel technique for improving diagnosis, prognosis, and treatment of cardiac disease.

Assessment of tissue structure: myocardial fiber imaging: The architecture of myocardial fibers^[97] is closely linked to the mechanical and electrical function of the heart. A tool for its assessment could be relevant in the diagnosis and risk stratification of many pathologic conditions, such as cardiomyopathies, myocardial fibrosis, and congenital heart diseases,^[71] as well as conduction delays.^[99]

SWI can determine the fiber direction in skeletal muscle^[77] and myocardium^[78,79] as tissue stiffness and—with this—SW speed is higher in the direction of the fibers than across. Elastic tensor imaging (ETI)^[78] has been proposed to quantify this anisotropy. SWs are induced by acoustic radiation force, and ultrafast ultrasound imaging is used to track the SW propagation in different directions and in different layers of the tissue (Figure 11).^[71,80] This method has been tested in vitro in porcine myocardial samples^[73] and in vivo in open-chest animals.^[78] Villemain showed that fractional anisotropy by SWI is different between healthy volunteers and patients with HCM.^[22] This initial approach is challenging, however, because it requires repetitive acquisitions during a mechanical rotation of the probe. 3D elastic tensor imaging (3D-ETI) based on 4D ultrafast shear wave elastography imaging was more recently introduced. It overcomes the limitation of the 2D method by providing a volumetric quantitative map of tissue elastic properties within less than 20 ms of acquisition for the entire imaged volume.^[79]

Figure 11

The link between myocardial structure and cardiac function may have important potential clinical applications; however, it remains vastly unexplored because of technical challenges.