THREE-DIMENSIONAL ECHOCARDIOGRAPHY IN FUNCTIONAL ASSESSMENT
Assessment of ventricular volumes and ejection fraction is important in patients with congenital and acquired heart disease. Despite intrinsic limitations, ejection fraction still is the most commonly used parameter for ventricular performance. Measurement of ventricular volumes is important in patients with different cardiac conditions such as aortic or pulmonary valve regurgitation. Two-dimensional (2D) echocardiography has been traditionally used for assessing LV and RV volumes and ejection fraction. For the LV, the two most commonly used methods are the biplane Simpson’s method of disks and the area-length method. Both methods depend on geometrical assumptions and use an ellipsoid LV model. In patients with eccentric LV remodelling and LV dilatation, the ventricle becomes more spherical and the LV shape may also be more variable in congenital heart defects. Two-dimensional volumetric assessment of the RV is even more complicated due to its complex geometrical shape, which is difficult to capture in a simplified mathematical formula. The availability of three-dimensional imaging overcomes some of the limitations of the 2D techniques and, especially for assessment of LV volumes and ejection fraction, 3D technology with automated postprocessing, is becoming the technique of choice. The development of 3D matrix probes and automated border detection techniques have made 3D techniques more easily accessible for clinical use.
From Reconstruction Techniques to Real-Time Three-Dimensional Imaging
The development of 3D echocardiography imaging was a logical but challenging technological development. Already in the 1970s initial attempts were made to record and display images of the heart in three dimensions. At that time, off-line 3D reconstruction from serial ECG-gated acquisitions of multiple 2D planes was performed either by freehand scanning or a mechanically driven transducer that sequentially recorded images at predefined intervals. With freehand scanning, a series of images were obtained by manually tilting the transducer along a fixed plane, and a spatial locator attached to the transducer translated the 3D spatial location into a Cartesian coordinate system. The main disadvantage of this approach was the relative bulk of the spatial locator device, which made transducer manipulation difficult, and extensive postprocessing was required. Freehand scanning used a mechanical transducer for obtaining serial images at predefined intervals either in parallel planes or by pivoting around a fixed axis in a rotational, fan-like manner. Because the intervals and angles between the 2D images were defined, a 3D coordinate system could be derived from the 2D images in which the volume was more uniformly sampled. This approach, although accurate, required long acquisition and postprocessing times. The quality of 3D reconstruction from 2D images depended on a number of factors, including the intrinsic quality of the 2D data set, the number of 2D images used, the ability to limit motion artifacts, adequate ECG, and respiratory gating. It also relied on the assumption that all planes are acquired at the same phase of the respiratory cycle to ensure identical shape and position of the heart within the chest. Finally, the need to shorten the acquisition times in order to minimize motion artifacts, made image quality often suboptimal. Later, the use of multiplane probes emerged as a readily available method to obtain rotational images at predefined angles around a fixed axis.
An important milestone in the history of 3D echocardiography was reached in the early 2000s, with the development of fully sampled matrix array transducers. These transducers allow the creation of real-time 3D images. The need for tedious multiplane acquisition was eliminated. This novel approach is based on real-time volumetric 3D imaging. This advance in 3D imaging saved computer interpolation of cross-sectional images and thus could avoid spatial motion artifacts. Real-time 3D echocardiography uses transducers containing arrays of piezoelectric elements capable of acquiring pyramidal data sets. It is now possible to depict cardiac motion at a sufficient frame rate during a single breath hold without the need for off-line reconstruction, thus eliminating motion artifacts known to adversely affect the reconstruction methodologies.
Real-time 3D echocardiography uses matrix array transducers with elements arranged in a grid fashion, typically containing more than 3000 imaging elements transmitting at 2 to 4 MHz for transthoracic imaging. The availability of a higher frequency probe (7 MHz) made it possible to obtain real-time 3D images with improved spatial resolution in younger children and infants. The matrix probes are computationally demanding, and to reduce the size of the connecting cable, miniaturized circuit boards are incorporated into the transducer, allowing partial beam forming within the transducer. More recent developments resulted in a smaller transducer footprint, improved side lobe suppression, and harmonic capabilities. These probes can acquire data sets at 360-degree focusing and have electronic steering for volumetric acquisition at satisfactory spatial resolution (voxel size, 0.5 3 0.5 3 0.6 mm) with acceptable temporal resolution (30 to 40 frames/s).
Real-time 3D systems generally have two methods for data acquisition: real-time, also called live 3D imaging, and electrocardiographically gated multi-beat 3D imaging. Real-time imaging refers to the acquisition of multiple pyramidal data sets over a single heartbeat. Real-time 3D data can be acquired in three modes: narrow-angle; zoom, magnified; and wide-angle. Although these modes overcome the limitations imposed by rhythm disturbances or breathing motion, the data sets are acquired at lower spatial and temporal resolution. The real-time narrow mode displays a pyramid of approximately 60 3 30 degrees that permits visualization of a single, relatively small structure like the aortic valve, in any one imaging plane with good spatial resolution. The zoom mode displays a magnified, wide pyramid of 30 3 30 degrees (Video 4.1). The wide-angle shows a pyramidal data set of approximately 90 3 90 degrees. Wide-angle data sets provide larger pyramidal scans at the cost of lower spatial resolution compared with narrow-angle acquisition.
Multi-beat 3D imaging provides images of higher temporal resolution through the acquisition of four to eight narrow volumes of data over several heartbeats that are subsequently stitched together to create a single volumetric data set (Video 4.2). Multi-beat imaging allows the acquisition of the largest sector possible at a reasonable temporal resolution of 30–40 frames per second. The full volume allows entire visualization of large structures such as the LV, RV, and valves. The full volume can be rotated to orient structures in face views (Fig. 4.1). These data sets can also be cropped or transected in any plane in order to better visualize specific anatomic structures (Fig. 4.2). Multi-beat volumetric imaging is inherently prone to artifacts created by irregular heart rhythm or motion (breathing). To minimize reconstruction artifacts, data should be acquired during breath holding if possible.
Recognizing and avoiding potential artifacts is critical for an accurate interpretation of 3D data. Artifacts are mainly related to respiratory motion in patients unable to hold their breath, like young children. Motion artifacts may result in stitching artifacts when the four sub-volumes are merged. ECG gating can be challenging in patients with arrhythmias, including normal sinus arrhythmia. Importantly, in real-time 3D echocardiography, image quality is closely related to the intrinsic 2D images used to produce the 3D data set. When the 2D images are poor then the 3D images are generally even poorer and should not be acquired. The use of optimal gain settings is essential for accurate diagnosis. Low gain settings can artificially fade certain structures that will not be viewable during postprocessing. In contrast, excess gain leads to a decreased spatial resolution and loss in the 3D perspective. It is generally recommended to use higher gain settings during acquisition with adjustment of the gain settings during postprocessing.
Figure 4.1. Three different three-dimensional data acquisition modes. A: Real-time narrow-angle imaging mode. B: Zoom mode, which allows visualization of small structures such as a coaptation abnormality of the tricuspid valve. C: Wide-angle full-volume data set; this data set is acquired over four to seven heartbeats.
Figure 4.2. A two-dimensional (2D) (A) and full-volume three-dimensional (3D) (B) data sets cropped to view cardiac structures of interest in a patient with left ventricular noncompaction. Compared with the 2D image, the 3D image shows depth information at the cost of lower spatial resolution.
In 3D imaging there is an important trade-off between spatial and temporal resolution (volume rates). To improve spatial resolution, an increase in the number of scan lines per volume is required; this, however, takes longer to acquire and reduces temporal resolution. Imaging volumes should be made as small as possible to increase temporal resolution while maintaining spatial resolution.
A complete 3D echocardiogram should include assessment of ventricular morphology, volumes and function, valve morphology, and hemodynamic data. In general, 3D echocardiography is performed as a complement to a 2D study. Although a 3D full-volume acquisition mode can accommodate almost the entire heart with existing technology, the decrease in spatial and temporal resolution that would result from enlarging the volume to acquire the entire heart from a single acoustic window makes this impractical. Therefore, 3D data sets are acquired from multiple transducer positions. For example, focused 3D imaging for LV volume quantification typically can be performed with an apical four-chamber wide-angle acquisition complementing standard 2D imaging (Fig. 4.3 and Video 4.3). An advantage of 3D echocardiography compared with 2D echocardiography is that quantitative analysis does not rely on the ability of the operator to acquire images positioned properly in certain standard views relative to cardiac anatomy. Unlike 2D echocardiography, in which standard views are described based on the plane through which they pass, 3D echocardiography is inherently volumetric. As such, full-volume data sets of the beating heart, which can be rotated and sliced in any plane, allow the reader both an external view of the heart and multiple internal views with the use of cropping (Video 4.4). Essential components of a full 3D echocardiogram for assessment of ventricular function include wide-angle acquisition of a parasternal long-axis view, an apical 4-chamber view, and a subcostal view including color interrogation of all valves. Visualization, description, and analysis of images are generally performed in three orthogonal planes: (a) the sagittal plane, which corresponds to a vertical long-axis view of the heart; (b) the coronal plane, which corresponds to a four-chamber view; and (c) the transverse plane, which corresponds to a short-axis view. Each plane can be viewed from two sides, representing opposite perspectives. For example, the transverse plane can be viewed from the apex or the base of the heart. The choice of narrow- or wide-angle acquisition depends on the structure to be visualized. For smaller structures, such as valves, a narrow-angle acquisition is more appropriate. For imaging the ventricles, it is best to use a wide-angle acquisition in the apical window so that the entire volume of the ventricles can be covered. Acquisition is followed by off-line analysis with dedicated 3D software. This requires cropping and tilting of the data set. Because a data set comprises the entire ventricular volume, multiple slices can be obtained from base to apex to evaluate wall motion. An advantage of 3D over 2D imaging is the possibility of manipulating the plane to align the true long and short axes of the left ventricle, avoiding foreshortening and oblique imaging planes.
Figure 4.3. A full-volume data set (bottom right) cropped into coronal, sagittal, and transverse planes. Such data sets are used to delineate endocardial and epicardial borders and calculate ventricular volumes, mass, and ejection fraction.
Assessment of LV Volumes and Ejection Fraction
Accurate and reproducible quantitative assessment of LV size and mass, as well as regional and global LV function, are important components of an echocardiogram. They are pivotal for diagnosis, treatment, and prognosis of patients with different heart diseases. Multiple methods for measuring LV size and function have been developed for both M-mode and 2D echocardiography. The relative inaccuracy of these one-dimensional (1D) and 2D approaches has been attributed to the need for geometric modeling of the ventricle, assuming the left ventricle is an ellipsoid. The lack of information on one dimension, the “missing dimension,” has been considered the main source of the wide inter-measurement variability of the echocardiographic estimates of LV size and function. This is particularly true in children with congenital heart disease whose ventricles have distorted morphology and do not follow geometric models that are generally used. An advantage of 3D echocardiography over 2D imaging is that it provides full-volumetric data sets, so there is no need for geometric modeling. Several studies comparing 3D echocardiography with magnetic resonance imaging (MRI) as a gold standard have shown that the quantification of LV volumes and function with 3D echocardiography is highly feasible, reproducible, and accurate when compared with LV volumes measured by MRI (Fig. 4.4).
Visualization of the endocardial surface can be challenging, especially in the apical lateral and anterior myocardial segments. This can be compensated for by tilting the transducer, which improves endocardial visualization at the expense of generating foreshortened views of the left ventricle, resulting in a potential source of error when calculating LV volumes. Despite the high correlation with MRI, studies have shown 3D echocardiography slightly underestimates LV volumes. This is believed to be due to the inability of 3D echocardiography to identify the myocardial borders as accurately as MRI. In addition, the lower temporal resolution of 3D echocardiography may lead to the inability to capture the real end-systolic frame. This can lead to inaccurate end-systolic volume measurements.
In the past, 3D echocardiographic quantification of LV volumes and function required tedious manual delineation of subendocardial borders in multiple planes. Today, most vendors offer a nearly fully automated frame-by-frame detection of the endocardial surface. Typically this process involves segmentation of the 3D data set into two or three equiangular 2D longitudinal planes. This requires indicating a few anatomical landmarks such as the septal, lateral, anterior, and inferior mitral annuli at end diastole and end systole. If necessary, manual corrections can be performed after automated analysis. The software calculates LV volumes in each frame, endocardial wall motion, and ejection fraction (with the use of no geometrical assumptions) by detecting the endocardial blood interface. Models of the left ventricle and data from normal LVs, however, guide border detection methods. The data sets can be displayed as a volume-rendered or surface-rendered image (Figs. 4.5 and 4.6 and Videos 4.5 and 4.6).
Figure 4.4. Multislice short-axis display of a full-volume data set. Cropping and rotation in any desired plane allows the operator to render the true short-axis views covering the entire left ventricle from base to apex. This display is useful to evaluate wall motion abnormalities because it allows direct comparison among different myocardial segments.
Three-dimensional echocardiography also allows calculation of LV mass. Two-dimensional echocardiography derives LV mass from measurements of myocardial wall thickness at one site and uses geometrical formulas. This leads to inaccuracies, especially when wall thickness is inhomogeneous, as occurs in asymmetric hypertrophic cardiomyopathy. Three-dimensional echocardiography seems to overcome this limitation because it allows calculation of wall thickness in all myocardial segments. Three-dimensional echocardiographic measurement of LV mass requires the delineation of endocardial as well as epicardial borders. While M-mode overestimates LV mass and 2D echocardiography underestimates LV mass, 3D echocardiographic measurements have been shown to correlate highly with MRI mass calculation.
The current analysis programs also allow calculation of regional myocardial function based on the 3D data sets. Changes in regional volumes can be calculated and regional volumetric curves can be displayed for each sub-volume. This allows quantification of regional myocardial function. Speckle-tracking technology can also be applied to the volumetric data sets to calculate 3D strain. A clinically useful byproduct of the 3D quantification of regional LV wall motion is the ability to quantify the temporal aspects of regional systolic wall thickening. The standard deviation of regional time to peak ejection times (interval between the R wave and peak systolic endocardial motion) has been used as an index of myocardial synchrony (Figs. 4.7 and 4.8 and Videos 4.7 and 4.8). This index can also be used to select candidates and to guide resynchronization therapy. A reduction in acquisition time makes 3D echocardiography particularly suited to stress echocardiography in which there is a narrow temporal window to acquire images at peak stress. The downside is the lower frame rates at higher heart rates. The benefit of this methodology in pediatric heart disease awaits further studies.
RV Volumes and Ejection Fraction
An accurate evaluation of RV size and function is of great importance in patients with congenital heart disease. However, imaging the RV is challenging due to its complex shape, its anterior position in the chest, and its thin walls with deep trabeculations. Mainly because of the more complete morphology of the RV walls, cardiac MRI is considered the clinical reference technique for measuring RV volumes and ejection fraction. Even when using cardiac MRI, standardization in image acquisition and postprocessing is essential for reducing interobserver variability. In younger children, cardiac MRI requires general anesthesia and is not readily available for serial measurements.
Figure 4.5. Calculation of left ventricular end-diastolic and end-systolic volumes, stroke volume, and ejection fraction in a 4-year-old with aortic stenosis and heart failure. Note that ventricular volumes are increased and ejection fraction is reduced. Also, significant heterogeneity in regional volumes indicating differences in regional myocardial function is seen (bottom left).
Figure 4.6. Analysis of left ventricular global and regional function, as well as global and regional volume curves with three-dimensional echocardiography. The endocardial border is semiautomatically traced after manual indication of the mitral annuli and apex in two orthogonal views (triangles) at end diastole and end systole.
In theory, 3D echocardiography could be a good alternative to cardiac MRI for the assessment of RV shape and function. This requires a specific acquisition of RV volumes to capture the different RV walls in the volumetric data set. Specific postprocessing software is available for analyzing RV volumes and ejection fraction. This still is a semi-automated method requiring more extensive postprocessing compared to LV volumetric analysis (Fig. 4.9). Another limitation is that good quality data sets are often difficult to obtain, as visualization of the entire right ventricle from a single data set is often not possible. It is not uncommon that a part of the RV outflow tract cannot be included in the data set. This is particularly problematic for the more dilated RV, limiting the feasibility of this method, especially in adult patients (Video 4.9).
Different studies have tried to validate this methodology in different patient groups and compared the volumetric 3D RV data with measurements obtained by cardiac MRI. While certain studies report reasonable accuracy, the majority of the studies report a systematic underestimation of RV volumes using the echocardiographic methods, especially when the RV is more significantly dilated. Some groups have tried to subsegment the RV volumes into inlet, apical body, and RV outlet.
An alternative method for quantifying RV volumes uses 2D images positioned in 3D space using a magnetic tracker attached to the probe. On the 2D images, anatomic landmarks are identified and these landmarks, localized in 3D space, are used for reconstructing the RV volumes based on a database of RV shapes. This method has the advantage of being less dependent on the acquisition of full-volumetric 3D data sets and, as based on 2D images, is more often successful in obtaining reliable RV data. It has also been shown to be highly reproducible, and the RV volumes obtained using this method correlate well with the volumes obtained by cardiac MRI. The disadvantage of this method is that it requires the use of a magnetic tracking device and specific software and the online access to the database of RV shapes. This limits the accessibility of the methodology.
Figure 4.7. Assessment of left ventricular synchrony in a normal subject. A and B: Ventricular volumes, ejection fraction, and synchronicity index calculated as standard deviation in time from R wave to minimal regional volume in 16 myocardial segments. Note small variation among myocardial segments. C: Regional myocardial volume curves throughout the cardiac cycle. D: Bull’s eye representation of the same data. Note homogeneous activation pattern with slightly more delayed activation of the basal septal and basal lateral myocardial segments.
Figure 4.8. Assessment of left ventricular synchrony in a child with dilated cardiomyopathy. Note enlarged left ventricular volumes with reduced ejection fraction and large standard deviation of time from R wave to minimal regional volume; this indicated dyssynchrony (Aand B). C: Bull’s eye representation demonstrating earlier activation of the interventricular septum and inferior wall with delayed activation of the basal anterior and lateral myocardial segments. D: Curves with marked dispersion of time to minimal regional volume.
Figure 4.9. Right ventricular analysis program displaying three-dimensional ultrasound images in sagittal (A), four-chamber (B), and coronal (C) views at end diastole. By manual tracing of endocardial borders at end diastole and end systole in all planes, one can calculate right ventricular volumes and ejection fraction (D).
Assessment Volumes and Ejection Fraction in the Single Ventricle
Assessment of single ventricular function remains a clinical challenge and in clinical practice is still largely based on subjective assessment. Having a more quantitative method available for clinical follow-up would certainly be useful. Full-volumetric acquisition with postprocessing using either LV or RV software depending on the morphology of the dominant ventricle could theoretically be used. There has been some validation of this methodology in various types of single ventricles. This earlier work was based on a method of disks, which requires extensive postprocessing and is no longer readily commercially available. Data on the more automated methods for assessing RV function are scarce. There are intrinsic limitations to the application of the 3D methodology to single ventricles. The most important one is that it can be very difficult or impossible to get the entire single ventricle in the volumetric data set. A second problem is that in case there is a smaller second ventricle also contributing to cardiac output, the analysis can become even more complicated. Finally, the identification of the different walls of the single ventricle can be difficult, limiting the feasibility and reliability of the method.
Additional potential applications of real-time 3D echocardiography in pediatric heart disease that are beyond the scope of this chapter include (a) measurement of left atrial volumes, which is considered to reflect chronic diastolic function and left atrial pressure; (b) quantification of ventricular volumes in the fetal heart; (c) visualization of valve morphology, calculation of valve area, and quantification of regurgitant volumes; (d) guidance of transcatheter interventions such as atrial septal defect device closure, electrophysiologic studies, and intraoperative monitoring; and (e) description of complex congenital heart lesions.
Future advances in transducer and computer technology should focus on higher frequency transducers for smaller children with a smaller footprint. Also, better temporal and spatial resolution of the 3D images would bring additional diagnostic benefit. Better 3D representation of the images and data sets through the development of 3D screens, holographic projection, and 3D printing techniques would allow better understanding of 3D anatomy. Finally, combining 3D echocardiography with MRI or computed tomography (CT) images may yield fusion data sets with unsurpassed anatomic, functional, and physiologic information.
TISSUE-DOPPLER IMAGING FOR THE EVALUATION OF MYOCARDIAL FUNCTION
Most of the traditional echocardiographic techniques used to assess ventricular function look at either dimensional changes caused by ventricular contraction or study the effect of cardiac events utilizing blood pool Doppler. All this indirectly reflects what is happening within the cardiac wall during the cardiac cycle. A different approach is to look directly into the myocardium and measure regional function by quantifying myocardial properties like velocities and deformation throughout the cardiac cycle. Tissue-Doppler imaging (TDI) was the first technique developed for studying myocardial motion. By adjusting machine filter settings, tissue velocities in the myocardium can be measured (Fig. 4.10). The initial description did not raise a lot of clinical interest; but when experimental data confirmed that, in normal myocardium, changes in segmental systolic velocities were linked to changes in regional contractility, there was a renewed interest in the measurement. Several clinical studies subsequently investigated the value of measuring regional myocardial velocities in various diseases such as ischemic heart disease, aortic insufficiency, and hypertrophic cardiomyopathy. In ischemic heart disease, it was demonstrated that tissue velocities changed very quickly (within 5 s) after the induction of myocardial ischemia in a pig model and was one of the earliest changes observed. Also for pediatric and congenital heart disease, this technique was considered to be potentially useful, as it directly looks into the myocardial walls and is independent of ventricular geometry. Myocardial velocities can be measured using pulsed-wave (PW) Doppler: a sample volume of 4–8 mm can be placed within the myocardium while trying to align the ultrasound beam as parallel as possible to the direction of motion studied by the technique (Fig. 4.11). Alignment is important, as all Doppler-based techniques are angle-dependent. The velocity waveform represents the instantaneous velocity in the region of interest throughout the cardiac cycle. PW Doppler has a very high temporal resolution (250–300 frames/s). The spatial resolution is limited because the sample volume does not track the translational motion of the underlying myocardial segment as the segment moves through the sample volume. A normal pulsed-wave tissue-Doppler imaging trace obtained in the basal part of the interventricular septum is shown in Figure 4.11. Different peaks can be noted. During the isovolumic contraction period, there is a short-lived peak corresponding to the myocardial shape change that occurs during this period. When the fibers contract against closed valves, there is a change from a more spherical to an ellipsoid shape. During the ejection phase there is a systolic waveform that can be measured that corresponds to the base-to-apex motion of the myocardium during systole. Normally this waveform peaks during the first one third of the cardiac cycle corresponding to the maximal force development early in systole. During the isovolumic relaxation period, another short-living velocity peak can often be recorded. In diastole two peaks are present: an early diastolic peak corresponding to early filling and a late diastolic peak occurring during atrial contraction. These velocities represent the opposite motion of the atrioventricular valve from the apex toward the base in diastole. As with all Doppler techniques, by convention, motion toward the transducer is represented as positive waves and motion away from the transducer as negative waves. In current clinical practice PW Doppler is used to quantify longitudinal myocardial motion by measuring mitral and tricuspid annular velocities. For the mitral annulus this can be measured in the septum or in the LV lateral wall. Peak septal velocities are lower compared to the LV lateral wall velocities. Peak tricuspid annular velocities are higher compared to mitral annular velocities.
Figure 4.10. Principles of tissue-Doppler. The tissue signals have higher amplitude but lower velocity while the blood pool signals have lower amplitude but higher velocity. By adjusting the threshold filter settings the tissue-Doppler velocity tracings can be detected.
Figure 4.11. Pulsed Doppler tracing obtained in the basal part of the interventricular septum. The pulsed-wave Doppler sample volume is placed in the basal part of the interventricular septum just below the mitral valve annulus. Different peak velocities can be measured: systolic peak velocity during ejection (S’), early diastolic peak velocity (E’) and diastolic peak velocity during atrial contraction (A’).
Tissue-Doppler velocities can also be measured using color tissue-Doppler imaging (CTDI). It was introduced in the early 1990s based on the same principles as color Doppler blood pool imaging. Using auto-correlation techniques, regional mean velocities instead of peak tissue velocities are measured. The difference in measurement technique explains why CTDI-derived myocardial velocities are on average 15–20% lower compared to PW Doppler myocardial velocities. The CTDI images are displayed as color images using the same color-coding used for blood pool color Doppler imaging with myocardial velocities toward the transducer displayed in red, and velocities away from the transducer in blue (Fig. 4.12). The color-coded tissue-Doppler information can be stored digitally and, by using postprocessing software, any velocity curve can be displayed at any point within the color Doppler information, allowing measurements at different sites in the myocardium during the same cardiac cycle. When narrowing the sector and optimizing the machine settings, high-frame rates to .250 fps can be obtained. This is important as certain myocardial mechanical events are short-lived and require high-frame rates. As with any Doppler technique, alignment with the direction of myocardial motion is important. Compared with PW TDI, CTDI has the advantage that velocities in different myocardial segments and walls can be recorded simultaneously during the same cardiac cycle. This allows comparing the regional wall motion and timing of cardiac events between different myocardial segments within the same cardiac cycle. This can be useful in the evaluation of the synchronicity of the contraction pattern in different myocardial wall segments and can also be used for calculating myocardial velocity gradients within a myocardial wall. This is the basis of TDI-based strain and strain rate imaging, which has largely been replaced by speckle-tracking echocardiography in clinical practice, but has been an important research technique.
Current Clinical Applications of Tissue Velocities
Multiple experimental and clinical studies have been performed to characterize the myocardial velocity profile in normal subjects. Systolic myocardial velocities have been used to look at systolic function, and in adults, measurement of peak early diastolic velocity (E’) has become a key component in the assessment of diastolic function, both for grading diastolic dysfunction as well as in the assessment of LV filling pressures. Velocities are vectors expressed relative to a coordinate system. In the cardiac coordinate system, motion is described relative to the longitudinal, radial, and circumferential direction (Fig. 4.13). This differs from an intrinsic coordinate system that ideally should be based on actual fiber orientation. In the epicardium, myofibers are predominantly obliquely and more longitudinally oriented in a left-handed helix. In the LV, the fiber orientation changes from epicardium to endocardium—from oblique to circumferential in the midwall into a right-handed obliquely oriented helix in the endocardium. The underlying fiber orientation influences the velocities measured in a certain direction or axis. In the longitudinal direction, systolic and diastolic myocardial velocities are higher at the base of the heart and decrease toward the apex, which remains almost stationary. Due to a predominant longitudinal motion in the RV, the tricuspid annular tissue-Doppler velocities are higher than the LV longitudinal velocities. When analyzing radial myocardial function, myocardial velocities are higher at the endocardium than at the epicardium. For these reasons, a gradient in myocardial velocities exists between the base and the apex (in the longitudinal direction), and between the endo- and epicardium. This velocity gradient can be measured and represents myocardial strain rate (the rate of deformation in a certain segment). When integrating the strain rate curve versus time, myocardial strain can be calculated. With the advent of speckle-tracking echocardiography, these techniques have become obsolete in clinical practice.
Figure 4.12. Color Doppler myocardial imaging. An apical four–chamber view is obtained with color tissue–Doppler imaging. From the color-velocity data, tissue velocity curves can be extracted and on these traces the same velocity peaks can be identified as on the pulsed-wave traces (S’, E’, and A’). The advantage of color Doppler myocardial imaging is that velocities in different segments can be recorded during the same cardiac cycle. This allows comparing velocities in different segments during the same cardiac cycle.
Figure 4.13. The cardiac coordinate system. This is the cardiac coordinate system used as a reference system for tissue-Doppler vector velocities and myocardial deformation imaging. Velocities and deformation are expressed in the longitudinal (L), radial (R), and circumferential (C) direction. Longitudinal velocities are measured from apical views; radial from short-axis or long-axis views; circumferential from short-axis views.
Use of Tissue-Doppler Velocities in Children
Normal values for PW tissue-Doppler velocities in children have been published in different studies. In children, myocardial velocities vary with age, heart rate, and myocardial segment. A study published by Eidem et al. included 325 children from different ages. They showed that pulsed-wave tissue-Doppler velocities are influenced by cardiac growth, especially by left ventricular (LV) end-diastolic dimension and LV mass. Similar findings have been reported during fetal growth with a progressive increase in tissue-Doppler velocities with increase in cardiac size. This indicates that tissue velocities are, to some degree, influenced by ventricular geometry, which has significant implications when applying this methodology to children with congenital heart disease. A possible solution to this problem has been proposed by Roberson et al. who expressed normal TDI velocities as z-scores normalized for BSA. In adult studies, peak tissue-Doppler velocities were initially reported to be relatively load-independent, while later studies could demonstrate the influence of loading conditions (preload and afterload) on peak systolic and early diastolic tissue velocities. A distinction has to be made between acute changes in loading conditions and chronic pressure or volume loading. Cardiac remodeling (eccentric or concentric hypertrophy) associated with chronic loading can result in normalization of loading conditions, which can also result in a normalization of the tissue-Doppler patterns. Acute changes in preload and afterload generally influence peak systolic tissue-Doppler velocities, while velocities may normalize again in chronically adapted ventricles. Eidem et al. studied the effect of different congenital abnormalities on tissue-Doppler velocities in children. In patients with dilated cardiomyopathy, systolic tissue velocities were reduced in the different segments, consistent with a reduced ventricular systolic function in this patient group (Fig. 4.14). In children with ventricular septal defects, basal septal velocities were only very mildly reduced with normal systolic and diastolic peak velocities in the LV basal lateral wall. In contrast, children with aortic valve stenosis had reduced systolic basal velocities that were reduced in both the septum and the lateral wall when compared to velocities in a normal control group. This suggests that longitudinal systolic function is decreased in patients with aortic stenosis. Similar observations were made by Kiraly et al. with decreased systolic and diastolic tissue velocities in the lateral and posterior LV walls with longitudinal velocities more significantly reduced compared to radial velocities in children with aortic stenosis. Weidemann et al. showed that in adult patients with aortic stenosis the degree of reduction in LV longitudinal function is predictive of the degree of myocardial fibrosis as observed during surgery. Data for children are not available, however.
Figure 4.14. Longitudinal systolic tissue velocities in patient with dilated cardiomyopathy. This 8-year-old child was diagnosed with dilated cardiomyopathy with globally reduced LV systolic function. The longitudinal tissue-Doppler velocities as obtained in the LV basal interventricular septum (A) and lateral wall (B) are significantly reduced.
In the assessment of RV function, assessment of longitudinal function is even more important because of the predominant longitudinal contraction in the normal RV. Peak systolic velocity at the tricuspid annulus (S’) is a good index of basal longitudinal RV performance but only represents global RV function if there are no regional wall motion abnormalities. Different studies have reported RV longitudinal velocities in children with different congenital heart defects. Eyskens et al. showed that RV systolic velocities were elevated in patients with atrial septal defects before closure of the defect and that these values normalized after percutaneous closure of the ASD. Pauliks et al. had similar findings in their study including children with atrial septal defects before and after interventional device closure. At baseline, children with ASDs had increased tricuspid and mitral annular velocities compared to controls while isovolumic RV acceleration, which is explained below, was similar between the two groups. Following ASD closure, a transient immediate decrease in tissue-Doppler velocities in all myocardial segments was demonstrated with tissue-Doppler velocities normalizing 24 hours postprocedure. Quantitative assessment of RV performance using tissue-Doppler after repair of tetralogy of Fallot (TOF) has also been the subject of different investigations. Tissue-Doppler velocities are decreased in tetralogy patients postrepair with some regional wall motion abnormalities detectable in the right ventricle (Fig. 4.15). In TOF patients, important regional dysfunction can be present with especially reduced function in the patched RV outflow tract that is often significantly dilated and aneurysmal. This explains why peak systolic tissue-Doppler velocities measured in the basal RV segment do not correlate well with global RV ejection fraction in patients with moderate to severe dysfunction in the RV outflow tract. Regional functional parameters should not be used as parameters for global function in any condition where significant regional wall abnormalities are present.
Loading conditions (pressure and volume loading) influence peak systolic velocities. Parameters measured during the isovolumic periods in the heart are generally less afterload-dependent. During the isovolumic contraction period a myocardial velocity spike can be detected. By measuring the acceleration or slope of this spike, myocardial acceleration during the isovolumic contraction period can be measured (IVA). IVA is calculated as the average rate of myocardial acceleration during the isovolumic contraction period expressed in cm/s2 (Fig. 4.16). Validation by invasive measurements of contractile function has suggested that IVA is a relatively load-independent parameter for contractile function. As isovolumic contraction is a short-lived event (30–40 ms), calculation of IVA requires obtaining images at the highest temporal resolution (.200 frames/s). The disadvantage is that IVA is highly heart rate–dependent which limits its use in baseline conditions. The heart rate dependency has been used to measure the force-frequency relationship during heart rate manipulation (stress echocardiography or pacing). Particularly during exercise, force-frequency relationships can be studied and normal responses in the pediatric age group have been described. In patients after tetralogy of Fallot, baseline IVA was reduced and the degree of dysfunction was found to correlate with the severity of pulmonary regurgitation. IVA has also been shown to be a useful noninvasive marker of allograft rejection in pediatric heart transplant patients, and a reduction in the IVA force-frequency relationship was found after cardiopulmonary bypass in children with different congenital defects. The value of IVA as a global load-independent functional parameter was questioned by Lyseggen et al. In their experimental animal study they showed that IVA is preload-dependent when left ventricular end-diastolic pressure is elevated. Moreover, there was no consistent relationship between regional IVA measurements and regional myocardial contractility in ischemic heart segments. IVA seems to be a poor parameter for regional function and should only be used as a parameter for global function when there is no significant regional dysfunction.
Figure 4.15. Peak systolic tissue-Doppler velocities in RV lateral wall segments in patients after tetralogy of Fallot repair. Longitudinal peak systolic tissue-Doppler velocities are reduced in the basal, mid, and apical RV free wall segments. (Based on Weidemann F, Eyskens B, Mertens L, et al. Quantification of regional right and left ventricular function by ultrasonic strain rate and strain indexes after surgical repair of tetralogy of Fallot. Am J Cardiol 2002;90:133–138.)
Figure 4.16. Measurement of isovolumic acceleration. On the tissue velocity tracing, the isovolumic velocity peak is identified. The mean acceleration is measured starting from the baseline to the maximal velocity. This measurement requires a high temporal resolution and high-frame rate as the isovolumic period is a short-lived event, especially at high heart rates. This also influences measurement errors.
Apart from the dependency on age, heart rate, geometry, and loading, tissue-Doppler velocities have additional limitations. When acquiring myocardial velocities, the motion of the heart within the chest (cardiac translation) is also measured. Additionally, segmental myocardial velocities are influenced not only by the intrinsic segmental function but also by adjacent myocardial segments (defined as tethering). The identification of regional myocardial dysfunction by tissue-Doppler echocardiography can be limited when regional disease is present (e.g., a localized myocardial infarction), as the diseased segment(s) may continue to move relatively normally due to the influence of healthy adjacent myocardium. As such they are not a true regional functional parameter as influenced by segmental interactions. Utilizing myocardial deformation imaging overcomes this limitation.
Deformation Imaging in Pediatric and Congenital Heart Disease
Strain Rate/Strain Imaging: The Principles
Myocardial strain and strain rate imaging measure regional myocardial deformation. Regional strain represents the amount of deformation or the fractional change in length or thickness in a myocardial segment and is dimensionless, although sometimes presented as a percentage (%). Strain rate represents the velocity of myocardial deformation and is expressed as /s or %/s (Fig. 4.17). Strain rate is considered a better parameter for myocardial contractile function, as it is less dependent on loading conditions. Strain rate, however, is technically more difficult to measure, as strain rate curves are dependent on high-frame rates and are generally noisier compared to strain curves. Two different echocardiographic techniques are currently available for measuring myocardial deformation: tissue-Doppler–derived strain and strain rate techniques and speckle-tracking echocardiography. The tissue-Doppler–based techniques were the first techniques used for the calculation of myocardial deformation. The concept is based on myocardial velocity gradients that are present between the base and apex in the longitudinal direction and between the endocardium and epicardium in the radial direction. Regional strain rate was calculated from the spatial velocity gradients between two neighboring points in the myocardium (Fig. 4.18). The underlying principle is that instantaneous differences in tissue velocity between two adjacent segments reflect either expansion or compression of the tissue in between. Regional strain rate is the rate of deformation (/s) and can be measured from the velocity difference between myocardial points, divided by the distance between them (5 area of computation). For pediatrics, computational distances of 4–5 mm in the radial direction and 8–9 mm in the longitudinal direction were typically used. This technique measures natural strain, which means the instantaneous deformation over an infinitesimal small time interval, in contrast to MRI tagging where Lagrangian strain (change in length compared to the original length) is measured. Deformation can be compression, i.e., shortening in the longitudinal direction and thickening in the radial direction in systole, or expansion, i.e., lengthening in the longitudinal direction (diastole) and thinning in the radial direction in diastole. Shortening is either active contraction or passive recoil after stretching. Similarly, elongation can be relaxation after contraction or passive stretching. By convention, compression is characterized by negative strain rate and strain values and expansion by positive values. In contrast to myocardial velocities, strain rate and strain are not influenced by global heart motion and motion in adjacent segments, and are therefore better indices of true regional myocardial function. The disadvantage of the tissue-Doppler based techniques is the angle-dependency, which limits the number of segments that can be measured (mainly longitudinal deformation from apical views and radial deformation in the posterior wall from the short-axis views). Measurement of circumferential and rotational mechanics was not possible or was technically challenging. A lot of the initial research on myocardial deformation was based on the tissue-Doppler techniques, but the introduction into routine clinical practice was difficult due to the extensive postprocessing and the relatively poor reproducibility. The introduction of speckle-tracking echocardiography (STE) for estimation of myocardial deformation made the clinical use of myocardial deformation imaging more accessible. STE is based on gray scale tracking of 2D “speckles,” which are ultrasonic reflector patterns within the myocardium. These speckles can be tracked as they move through the cardiac cycle to determine the degree of deformation (Fig. 4.19). The advantage is that this method is angle-independent and requires less postprocessing. This method also allows calculating two-dimensional deformation parameters and the methodology reliably estimates longitudinal and circumferential myocardial deformation but less reliably measures radial deformation. These methods also allow measuring rotational mechanics and calculate LV torsion and twisting (Fig. 4.20). Rotation (expressed as degrees) is the relative clockwise or counterclockwise motion of the LV around the long axis of the heart when viewed in the short axis from the apex to the base. By convention, counterclockwise rotation is displayed as a positive value and clockwise rotation as negative. At the base there is a normal early counterclockwise rotation followed by a more dominant clockwise rotation, while at the apex the rotation is counterclockwise. The net difference between apical and basal rotation can be calculated and represents LV twist. LV torsion is the base-to-apex gradient in rotation angle corrected for LV length (degrees/cm).
Figure 4.17. Strain and strain rate definitions. The concept of strain and strain rate are illustrated on the M-mode through the LV walls. In the LV posterior wall the % thickening at end systole represents systolic radial strain in this segment. It is the percent thickening observed at end systole. Strain rate represents the rate of myocardial deformation. It is analogous to measuring the slope of the posterior wall thickening from baseline to peak, which would represent an average radial strain rate.
Figure 4.18. Radial and longitudinal strain and strain rate from tissue-Doppler data. In left panel, radial myocardial deformation estimation from the inferolateral wall is illustrated. From a short-axis view containing myocardial velocity data, the radial strain rate is calculated. Wall thickening in systole is displayed as positive values, whereas wall thinning in diastole is displayed as negative values. Temporal integration of the strain rate curve results in the local strain profile. Note strain rate peaks early in systole whereas strain peaks late in systole. Dotted lines indicate aortic valve closure. In the right panel, longitudinal myocardial deformation estimation in the interventricular septum from apical 4-chamber view is illustrated. Shortening in systole is displayed as negative values, whereas lengthening in diastole is displayed as positive values.
Figure 4.19. Principles of speckle-tracking echocardiography. In speckle tracking, speckle formations in gray-scale echocardiographic images are used as tissue markers that can be tracked from frame to frame throughout the cardiac cycle. Relative changes in position of these speckles are used to quantify regional deformation in a more automated fashion.
The disadvantages of speckle tracking include lower frame rates, which especially impact estimation of peak strain rate measurements. Another important limitation is that different commercially available software packages employ different postprocessing algorithms with different smoothing techniques, which influence the measurements and results in differences between the different software solutions. These software packages are functioning as “black boxes,” generating output from the two-dimensional images with limited input from the user. Standardization between the different vendors is required and currently is a high priority for the vendors. More recently, three-dimensional strain technology has become available which allows a full 3D evaluation of ventricular mechanics based on full-volumetric acquisition. The technology still has relatively low frame rates, and further technical optimization is required before it can be applied in routine pediatric use.
Figure 4.20. Rotation and twist. Speckle-tracking echocardiography allows quantifying ventricular rotational mechanics. When viewed from the apex, the base of the heart rotates clockwise (by convention a negative value) while the apex rotates counterclockwise (by convention a positive value). The difference between apical and basal rotation is the LV twist.
When analyzing regional function, it is always important to compare the local measurements to global timing events, because when inhomogeneities in contraction pattern are present within the heart (bundle branch block, ischemia, asynchronous contraction pattern), a significant amount of thickening or shortening can occur after aortic valve closure (postsystolic deformation). Therefore timing of mechanical events (aortic valve closure and mitral valve closure) is important to recognize the presence of postsystolic events. An example is given in Fig. 4.21, in a patient with hypertrophic cardiomyopathy. In the basal part of the ventricular septum, there is severely reduced strain and a significant amount of postsystolic shortening. In the mid-apical part, which is less hypertrophied, deformation is higher and there is much less postsystolic shortening. Postsystolic events occur when there are regional differences in myocardial function between different segments. This is the case in patients with ischemic heart disease or any disorder where regional differences in function can be found.
Strain Rate/Strain Imaging: Experimental Validation
Urheim et al. validated Doppler myocardial strain imaging as a new method to quantify regional myocardial function. They compared sonomicrometry with ultrasound measurements in various conditions including volume loading and coronary occlusion and demonstrated that the Doppler-derived strain values approximated those measured by sonomicrometry. Volume loading increased systolic strain and strain rate, which implied that both parameters are load-dependent. During coronary artery occlusion systolic myocardial velocities decreased in the non-ischemic segments, whereas regional strain and strain rate values remained unchanged. This suggests that Doppler-derived strain and strain rate are more direct measures of true regional myocardial function than tissue velocities, which are influenced by contractile function of adjacent segments due to tethering. In an experimental study in normal porcine myocardium, Weidemann et al. studied radial systolic strain rate and strain during alterations in inotropic states and heart rates. Strain is load-dependent, decreases with increasing heart rate and is correlated best with stroke volume. Strain rate is relatively independent of heart rate and better reflects contractility. An experimental dog study performed by Greenberg et al. demonstrated a strong correlation between systolic strain rate and maximal elastance, used as the gold standard for global contractility. Correlation between systolic velocities and elastance was weaker. These findings suggest that echo-Doppler–derived strain rate can be used as a noninvasive index of contractility. Ferferieva et al. demonstrated in invasive mice experiments that end-systolic strain measurements are highly afterload-dependent, while peak systolic strain rate is less influenced by acute changes in afterload. Jamal et al. validated RV systolic strain measurements against sonomicrometry and suggested that RV longitudinal end-systolic strain can be a good parameter for RV function. Two-dimensional speckle-tracking echocardiography was also validated against sonomicrometry and cardiac magnetic resonance imaging.
Figure 4.21. Postsystolic shortening in hypertrophic cardiomyopathy. Deformation pattern in the interventricular septum in a patient with asymmetric septal hypertrophic cardiomyopathy. Severely reduced systolic strain (also with abnormal systolic lengthening) and presence of postsystolic shortening occurs after aortic valve closure in the basal septum (arrows). The mid-apical septal myocardial segment shows relatively preserved systolic deformation and no postsystolic shortening. AVC: Aortic valve closure. (From Ganame J, Mertens L, Eidem BW, et al. Regional myocardial deformation in children with hypertrophic cardiomyopathy: morphological and clinical correlations. Eur Heart J 2007;28:2886–2894.)
Normative Myocardial Deformation Data in Children
Before clinical application in pediatrics, normal reference values had to be established. The first paper on normal tissue-Doppler derived strain rate and strain data in children was published by Weidemann et al. These data were based on a small group of only 33 healthy children (age 4–16 years). Boettler et al. obtained normal tissue-Doppler–derived strain measurements in different age groups. They observed that strain and strain rate values are influenced by heart rate, which was not validated in subsequent reports by other groups. Lorch et al. used Velocity Vector Imaging (VVI) (Siemens, Germany) and published normal values for longitudinal strain in 284 children. The normal range reported in this study is relatively wide, which probably reflects the high measurement variability. This limits the application of this technology in clinical practice. Marcus et al. used two-dimensional strain and published normal values for longitudinal, radial, and circumferential data in 139 children and 56 young adults using the Vivid-7 system (GE Healthcare, USA). In this study an effect of age on strain measurements was suggested with the highest values measured during the teenage years (15–19 years) and lower values seen in infants and in the adult group (.30 years). Normal pediatric data for LV rotation, torsion, and twisting were also recently published. Takahashi et al. included 111 normal subjects between 3–40 years (68 patients, 24 years). These authors could obtain full data sets in 66% of the normal subjects, demonstrating a more limited feasibility for rotational mechanics using 2D STE. The authors confirmed earlier data obtained with tissue-Doppler–derived techniques showing that with increasing age, net twist increases with age mainly as a result of a progressive increase in apical counterclockwise rotation. However, when the net twist was corrected for LV length and torsion was calculated, this was shown to remain constant across the age groups. In the neonatal age group, normal data were published using tissue-Doppler–derived strain analysis. Pena et al. calculated normal values for 55 infants within the first few days of life and then reassessed these data at 1 month of age. Only limited normal data for STE are available for this very young population, in part due to the technical difficulties of high heart rate, a small myocardial area, reduced number of speckles, and high degrees of artifacts, making adequate tracking for analysis problematic. Data on fetal strain are available but the methodology still requires further optimization and further clinical validation. Van Mieghem et al. applied this technology to study cardiac function in recipient fetuses in twin-to-twin transfusion syndrome and demonstrated a relatively low feasibility of the technique with data that could be obtained in only 61% of the included fetuses. A mild decrease in RV strain was demonstrated but the clinical significance of this finding remains unknown. Further optimization of the fetal speckle-tracking techniques is required before they can be implemented in routine clinical practice. New insights in fetal cardiac mechanics are likely going to be obtained using this methodology.
Clinical Applications of Myocardial Deformation Imaging in Pediatric Heart Disease
Since STE has become clinically available there have been a large number of studies published, including pediatric patients as well as adults with congenital heart disease. This methodology quickly replaced tissue-Doppler–derived myocardial strain calculations, which are technically more challenging, more variable, and have more limited scope. Still, a lot of the initial research on myocardial deformation imaging was performed using this technology, which had the advantage for pediatric usage of having higher temporal resolution compared with STE.
Strain imaging allows a reliable and highly reproducible quantification of LV and RV longitudinal function. It adds important additional information on myocardial function to the calculation of fractional shortening or ejection fraction.
Detection of Regional Myocardial Dysfunction
In adult acquired cardiac disease, strain imaging was introduced as a technique to quantify regional myocardial function in patients with ischemic heart disease. In coronary heart disease the evaluation of regional function is important. The most commonly used technique still is subjective scoring of regional wall motion. This requires extensive training and expertise, especially to recognize subtler wall motion abnormalities or milder forms of hypokinesia. Speckle-tracking echocardiography has the advantage of more objective quantification and allows detecting milder abnormalities in regional function like the areas of hypokinesia that can surround an akinetic infarcted segment. Regional myocardial dysfunction can also be present in children, related to congenital coronary artery anomalies after coronary artery transfer (i.e., after the arterial switch operation), or due to acquired coronary artery disease like Kawasaki disease or transplant vasculopathy. Pediatric cardiologists generally are less experienced with coronary artery problems compared with adult cardiologists and assessment of regional myocardial function is not as routinely performed as in adult echocardiography. Therefore tools for quantification of regional myocardial function are helpful in routine clinical practice for these patients (Fig. 4.22). An example of how deformation imaging gives additional information is illustrated in different conditions where global ventricular function is preserved but regional functional abnormalities can be detected by strain imaging. A good example is patients who are long-term after surgical repair for anomalous left coronary artery from the pulmonary artery (ALCAPA). Patients with a normal ejection fraction after surgical repair have been shown to have reduced longitudinal LV function as suggested by reduced mitral ring motion as well as reduced longitudinal deformation. The deformation data suggest that while radial function normalizes after coronary re-implantation, LV longitudinal function remains reduced, probably related to endomyocardial fibrosis mainly in the more vulnerable subendocardial region (Fig. 4.23). Deformation imaging has also been used for studying LV regional function in other disease conditions including hypertrophic cardiomyopathy, aortic stenosis, and coarctation of the aorta. The prognostic significance of these findings needs further investigation.
Figure 4.22. Regional myocardial dysfunction in transplant vasculopathy patient detected during routine follow-up. This 11-year-old patient was seen in the echocardiography laboratory 10 years after heart transplant for routine clinical follow-up. The patient had no symptoms and their ejection fraction was normal as calculated by different methods (M-mode, 2D biplane Simpsons, and 3D). Regional functional analysis by STE, however, revealed important hypokinesia in the lateral and inferolateral segments. Coronary angiography revealed important transplant vasculopathy.
Figure 4.23. Reduced longitudinal deformation in postoperative ALCAPA patients. Left ventricular longitudinal myocardial deformation at three levels in a normal control and a patient after coronary re-implantation for ALCAPA. The latter shows reduced longitudinal shortening.
Early Detection of LV Dysfunction
In patients at risk for developing ventricular dysfunction, ejection fraction or fractional shortening is generally the echocardiographic parameter used in general clinical practice. However, changes in EF occur only late in a chronic disease process due to the presence of different compensatory mechanisms to maintain cardiac output and stroke volume in the early stages of a progressive disease process. Myocardial deformation techniques were shown to identify detrimental changes in cardiac function prior to changes in ejection fraction. This has been demonstrated in patients exposed to cardiotoxic medication and in patients with genetic disorders associated with a risk for developing progressive LV dysfunction. A typical example is patients with Duchenne muscular dystrophy (DMD) who are born with mutations causing absence of dystrophin expression. This leads to a progressive dilated cardiomyopathy in a significant number of these patients later in life. In a significant number of young DMD patients the ejection fraction is within the normal range but myocardial deformation is reduced, especially in the inferolateral segments. Early changes in deformation parameters with preserved ejection fraction have also been reported in children with mitochondrial disease, Friedreich’s ataxia, and Prader-Willi syndrome. Abnormalities in myocardial strain parameters have also been described in obese children with known lipid abnormalities and in children with diabetes. The prognostic value of these findings is still uncertain and will require further longitudinal follow-up data.
An important population of pediatric patients that are being monitored for the development of LV dysfunction are those exposed to chemotherapy and radiation therapy. In chronic survivors, data obtained by tissue-Doppler–derived strain measurements as well as by speckle-tracking echocardiography showed mildly reduced circumferential and longitudinal strain parameters in adolescent patients who have been exposed to chemotherapy. These changes correlate with cumulative anthracycline doses. Rotational dynamics are also abnormal in childhood cancer survivors. Cheung et al. evaluated LV rotation in childhood leukemia survivors treated with anthracycline therapy. Although there was a reduction in ejection fraction compared to normal controls, there was also a subgroup of patients who had a normal ejection fraction but a reduction in peak torsion, apical untwisting, and LV systolic twisting velocity. All these findings strongly indicate that abnormal strain measurements reflect early changes in myocardial function prior to a decline in ejection fraction. This is suggested by studies performed in the more acute phase of the disease. In women treated for breast cancer with the cardiotoxic combination of anthracyclines and trastuzumab, the reduction in longitudinal strain at three months was shown to be predictive for a reduction in EF at 6 months. In children, Poterucha et al. found significant changes in longitudinal strain four months after starting chemotherapy, while significant changes in EF could only be detected at 8 months. Based on these initial data, strain imaging has been introduced in clinical follow-up in this patient population.
Also, for transplant graft surveillance in pediatric patients, data suggest deformation imaging may be a useful technique to detect early changes. Kailin et al. found a reduction in longitudinal strain but preservation of circumferential strain at 12 months after transplantation. In adult patients, Sarvari et al. showed that global longitudinal strain was an independent predictor of 1-year mortality posttransplant (p = 0.02). In addition, Marciniak et al. also showed that strain monitoring might be useful in myocardial acute rejection monitoring. The same group also suggested that tissue-Doppler–derived strain imaging was more accurate in detecting cardiac allograft vasculopathy using quantitative dobutamine stress echocardiography when compared to subjective visual assessment.
Regional Right Ventricular Deformation in Congenital Heart Disease
One of the most important challenges in congenital heart disease is the assessment of RV function. RV function is often very important, as in postoperative tetralogy of Fallot patients, patients after Senning or Mustard operations for transposition of the great arteries, patients with hypoplastic left heart syndrome, and other conditions. As the RV contraction pattern is more based on longitudinal shortening, studying RV longitudinal function is an important part of the assessment. In the systemic right ventricle, a change from more longitudinal to circumferential shortening and radial thickening has been observed, but is still more difficult to reliably assess in clinical practice. For clinical purposes a lot of attention has been focused on longitudinal RV function using myocardial deformation imaging. Longitudinal RV strain and strain rate can be calculated. In patients after tetralogy of Fallot repair, basal, mid-, and apical segments of the RV free wall peak systolic strain and strain rate values (Fig. 4.24) are reduced, and the degree of reduction in RV strain correlated with the degree of pulmonary regurgitation and also with exercise capacity. Important differences in regional myocardial function seem to exist within the RV free wall with more significantly reduced RV deformation in the more apical parts of the RV. This is very different from patients with atrial septal defects, where the apex seems to have higher regional strain compared with normal controls. Further study is required on the clinical significance of these findings. Another important insight is that there is a significant interaction between the right and left ventricle in tetralogy of Fallot patients and that progressive RV dilatation and dysfunction is related to LV dysfunction. Geva et al. demonstrated that RV and LV ejection fraction as measured by MRI highly correlate, and that a decrease in LV EF is an important predictor for clinical status and outcomes. Additionally, a reduction in LV longitudinal strain values was shown to be predictive for clinical events in this patient population, indicating that deformation analysis can play a role in predicting outcomes for this population. Recent studies also demonstrated that a reduction in RV longitudinal strain was a strong independent predictor for outcomes in patients with pulmonary hypertension, supporting a clinical role for this technology.
Figure 4.24. Right ventricular longitudinal strain analysis in patient after tetralogy of Fallot repair with pulmonary regurgitation. In this patient 12 years after tetralogy of Fallot repair, the right ventricle is significantly dilated related to severe pulmonary regurgitation. RV longitudinal strain is calculated using STE in the RV lateral free wall. While the basal longitudinal deformation is relatively well preserved and only mildly reduced, the RV apex is more significantly reduced when compared to normal RV values.
Also in patients with a systemic right ventricle, such as after Senning or Mustard repair for transposition of the great arteries, measurement of RV longitudinal strain can be performed. In this patient population, regional peak systolic strain and strain rate values were reduced in the basal, mid-, and apical segments of the RV free wall when compared with RV values in normal controls. The peak systolic strain values measured in the RV basal septum correlated highly with EF measurements obtained using cardiac MRI. This suggests that in this particular patient population deformation imaging could be used for serial follow-up of the patients. Similarly, reduced longitudinal RV deformation was measured in patients with congenitally corrected transposition of the great arteries. Also in the single RV, such as in patients with hypoplastic left heart syndrome (HLHS), this methodology can be used. Khoo et al. suggested that the RV in HLHS adapts by switching from a more longitudinal to circumferential contraction pattern and behaves more like a left ventricle. Lack of this adaptation may be associated with poorer outcomes.
Assessment of Mechanical Dyssynchrony
The identification of mechanical dyssynchrony in patients with congenital heart disease and ventricular dysfunction can be helpful for the decision for cardiac resynchronization therapy in addition to ECG criteria only. Initially, timing of myocardial events using tissue-Doppler velocities was thought to be a useful technique, but the disappointing results of the PROSPECT trial have tempered the initial enthusiasm for this methodology. During recent years much more attention has been given to the identification of patterns of mechanical activation. A typical example is left bundle branch block that can be characterized by early septal activation and septal shortening with late activation of the LV lateral wall. The early septal shortening will cause lengthening of the opposing LV lateral wall which is activated later. When the lateral wall is activated it stretches the septum, again causing the typical septal motion in this condition (early shortening followed by early stretching), also called the “septal flash” (Fig. 4.25). The occurrence of this pattern was highly predictive for the response to cardiac resynchronization therapy.
Figure 4.25. Abnormal septal motion in patients with left bundle branch block and left ventricular electromechanical dyssynchrony. Color M-mode is obtained based on parasternal short-axis images. In the left panel, the normal septal motion is shown, while in the right panel the abnormal motion of the septum with early contraction and early stretch the moment the opposing wall is activated. This creates the characteristic “septal flash.”
MAGNETIC RESONANCE IMAGING FOR THE ASSESSMENT OF VENTRICULAR FUNCTION
Echocardiography is the first-line noninvasive imaging tool in children with congenital and acquired heart disease because it is capable of providing comprehensive anatomic and hemodynamic information in many patients. However, with difficult acoustic windows, image resolution and its high dependency on operator experience limit its diagnostic utility. Cardiovascular MRI (cMRI) allows imaging of the entire thorax regardless of anatomic variations. For this reason, cMRI has become an established imaging modality to assess anatomy, physiology, and function in children with heart disease. cMRI can depict intra- and extracardiac anatomy in three dimensions, allowing quantification of ventricular volumes and ejection fraction and calculating flows. This section will review the role, advantages, and limitations of cardiac MRI in the assessment of ventricular function in children with heart disease.
Cardiac Magnetic Resonance Techniques Used for the Assessment of Ventricular Function
Three cardiac MRI techniques are most commonly used to assess cardiac function in patients with heart disease.
1. Cine MRI
This gradient-echo, steady-state free precession (SSFP), ECG-gated cine sequence yields images with high spatial and temporal resolution, excellent blood-to-myocardium contrast, and short acquisition time without the use of contrast agents (Fig. 4.26 and Video 4.10). Fast acquisition enables multiple phases of the cardiac cycle to be acquired. These can then be reconstructed into a cine MR image representing one full cardiac cycle using retrospective gating techniques at a temporal resolution of 20 to 25 ms. Multiple, contiguous cross-sectional, multiphase (cine loop) 2D slices are obtained across the region of interest to yield a spatially defined 3D data set at multiple levels and phases.
Cine MRI also allows evaluation of cardiovascular anatomy and function. Cine MRI is unique in that it creates images over several heartbeats, thus averaging ventricular performance in the process. This is in contrast to images obtained with echocardiography, in which each image instantaneously represents ventricular performance. The high spatial resolution allows delineation of the endocardial and epicardial borders to calculate ejection fraction, ventricular mass, stroke volume, cardiac output, wall thickness, and wall thickening. The SSFP sequence is, however, relatively insensitive to flow disturbances and highly sensitive to inhomogeneities in the field of view and metallic artifacts. Alternatively, the older spoiled-gradient technique (with lower contrast) can be used when delineation of abnormal flow jets is desirable or when implanted metallic devices produce significant imaging artifacts (Video 4.11).
Figure 4.26. Balance steady-state free precession (b-SSFP) since magnetic resonance imaging in short-axis view in a patient after repair of tetralogy of Fallot. The right ventricle is dilated, and the right ventricular outflow tract is thinned and dyskinetic.
Recently, a 3D SSFP technique has been developed. Fast imaging of the entire cardiac volume at slightly reduced spatiotemporal resolution but almost isotropic voxel size allows multiplanar reformatting. This is particularly useful in the assessment of complex congenital heart disease as it allows off-line reconstruction of complicated anatomy in any desired plane.
2. Velocity-Encoded Phase-Contrast MRI
Quantification of blood flow and velocity can provide relevant information for the management of patients with congenital heart disease. Velocity-encoded phase-contrast MRI uses phase information to encode velocity. In the images parallel to blood flow (in-plane), one can measure peak flow velocity. In the images perpendicular to flow (through-plane), the product of the velocity in each pixel by the area of the blood vessel will yield flow at that given period of time (Fig. 4.27). The addition of flow information across all phases of the cardiac cycle will yield flow during one heartbeat (Video 4.12). Direct quantification of stroke volume, flow, pulmonary-to-systemic ratio (Qp/ Qs), valvular regurgitation fraction, severity of valvular stenosis, and differential lung perfusion can be obtained with velocity-encoded phase-contrast MRI.
It is also possible to assess mitral and tricuspid inflow and pulmonary venous patterns to evaluate diastolic function. Because the MRI plane can be oriented in any direction, cardiac MRI avoids the difficulty of Doppler angle-dependency. This method has been validated against Doppler echocardiography. The disadvantage is the lower temporal resolution. Current indications of MRI include quantification of pulmonary regurgitation after repair of tetralogy of Fallot, quantification of aortic regurgitation, measurement of transaortic valve gradient when difficult by echocardiography, and quantification of (re)coarctation severity.
3. Myocardial Tagging
MRI is unique among imaging modalities in its ability to magnetically tag tissue. This is accomplished by applying thin saturation pulses immediately after the R wave. These saturation pulses destroy all the spins in a given plane, resulting in a line of signal void in the image. Similarly, two sets of orthogonal tags can be placed to produce a grid across the image (Video 4.13). This is followed by a standard but lower spatial resolution cine MRI sequence dividing the myocardial wall into cubes of magnetization. These lines or grids are distorted by myocardial motion, rotation, and deformation throughout the cardiac cycle (Fig. 4.28). Images can be obtained as frequently as every 15 to 20 ms in a given scan, yielding reasonable temporal resolution. The trade-off for this high temporal resolution is that the stripes tend to degrade and blur, so the assessment of myocardial wall mechanics in late diastole becomes less accurate. To perform calculations with tagged images, the initial step is to track the grid intersections through the phases of interest. This can be done manually or, as more recently described, semiautomatically. Tracking the translation, rotation, and deformation of these cubes in 2D or 3D allows calculation of wall motion, regional radii of curvature, regional wall thickening, and shortening as indicators of myocardial performance. Of course, the myocardial regions should be divided into anatomic regions (septal, anterior, lateral, and inferior walls) to perform the analysis. Full 3D analysis of circumferential, radial, longitudinal, and shearing myocardial forces is possible by collectively modeling numerous smaller elements (Fig. 4.29). This has led to a better understanding of myocardial mechanics. Through-plane motion, which echocardiography is not able to take into account, can be compensated for by myocardial tagging. Despite all the above-mentioned advantages, myocardial tagging has remained a research tool as a result of the lack of automated postprocessing software. Newer techniques such as harmonic phase imaging (HARP) and featured tracking appear promising because they do not require manual tracing of the tags; therefore, this shortens the analysis time significantly.
Figure 4.27. Phase encoded velocity mapping used to quantify flow and velocities of flow. A: Forward systolic flow in the right ventricular outflow tract in the cranial direction is displayed as white pixels. B: Flow velocity curve is obtained by measuring the flow profile over time.
Figure 4.28. Myocardial tagging creates a grid on the myocardium. Example of short-axis myocardial tagging at end diastole (A) and at end systole (B) in a patient with congenital aortic stenosis. At end systole, the original grid is distorted by myocardial deformation. Myocardial deformation is larger in the subendocardial than in the subepicardial layers. Note that the grids on the chest wall remain unchanged.
Figure 4.29. Analysis of circumferential myocardial deformation in a healthy 14-year-old boy. A short-axis image is used to perform analysis in four myocardial walls: septal (1), inferior (2), lateral (3), and anterior (4). Note that there is a homogeneous amount and timing of systolic myocardial deformation with rapid early diastolic recoil (arrow) in all myocardial segments.
In the clinical scenario, myocardial tagging has provided useful information in patients with ischemic heart disease. It has been shown that strain analysis can discriminate viable from necrotic myocardium. In patients with congenital heart disease, Fogel et al. have used myocardial tagging to characterize the pattern of wall motion and deformation in patients with functionally single ventricles. Myocardial tagging has demonstrated a shift in the systemic right ventricle pattern of contraction from predominantly longitudinal shortening of the subpulmonary RV to predominantly circumferential shortening with loss of torsion in the systemic RV.
Cardiac Magnetic Resonance to Assess Ventricular Function
Quantitative assessment of ventricular dimensions and function is an important part of MRI evaluation of children with heart disease because the evaluation of cardiac function provides valuable diagnostic and prognostic information. Cardiac MRI has the advantage of being a volumetric technique allowing 3D reconstructions. Cardiac MRI provides highly accurate and reproducible measurements of ventricular mass, volume, and function that have made it the reference standard against which other techniques are measured (Fig. 4.30). This makes cardiac MRI an ideal technique to follow up patients with heart disease and to monitor the response to therapies, because minor changes in volume/function can be detected with high accuracy. Normal MRI values of ventricular volume and mass in adolescents and adults normalized to body surface area have been extensively reported. Unfortunately, data on adjusting MRI-derived parameters to body size in the pediatric population come from very small studies (Table 4.1). Cardiac MRI is noninvasive and does not use ionizing radiation. It produces 3D data sets with high signal-to-noise ratio, high spatial resolution, and reasonable temporal resolution. This allows for accurate volume measurements of any cardiac chamber regardless of its morphology and without geometric assumptions. Independence from geometric assumptions is particularly important in patients with congenital heart disease, because the ventricles often have complex shapes that do not follow geometric assumptions used in formulas to calculate volumes from 2D data sets.
Figure 4.30. Evaluation of ventricular function, volume, and mass. A: Using a localizing image in the axial plane, a two-chamber (also known as vertical long-axis) plane is prescribed. B: From the two-chamber image, the four-chamber (or horizontal long-axis) plane is defined. C: From the four-chamber image, a 12-slice short-axis stack covering the ventricle entirely is prescribed. D: Resulting short-axis stack showing end-diastolic phase.
Quantitative evaluation of biventricular function can be achieved by obtaining a series of contiguous cine slices that cover the ventricles in the short-axis plane (Video 4.14). By tracing the endocardial border, the slice volume is calculated as the product of its cross-sectional area and thickness. Then ventricular volume is determined based on Simpson’s rule by simply adding the volumes of all slices. The process can be repeated for each frame in the cardiac cycle to obtain a continuous time-volume loop or may more simply be performed on only an end-diastolic and end-systolic frame to calculate end-diastolic and end-systolic volumes. Subtracting the end-systolic volume from the end-diastolic volume yields the stroke volume. The stroke volume divided by the end-diastolic volume gives the ejection fraction. Because the patient’s heart rate at the time of image acquisition is known, one can calculate LV and RV output. Ventricular mass is calculated by tracing the epicardial borders, subtracting the endocardial volumes, and multiplying the resultant muscle volume by the specific gravity of the myocardium (1.05 g/cm3).
Gradient-echo cine MRI is the technique of choice to measure RV volumes and function. RV volumes and function, as well as pulmonary flow, can be measured by MRI with a high rate of technical success in comparison with echocardiography. cMRI can directly measure RV volumes without the use of any geometric assumptions regardless of the position of the RV. At present, MRI quantification of RV volumes is considered the clinical reference technique for the evaluation of RV function. Table 4.2 shows the strengths and limitations of cardiac MRI compared with 2D and 3D echocardiography for the assessment of ventricular function.
Assessment of the contribution of different areas of the ventricular wall to global ventricular performance is important in many cardiac diseases. With cardiac MRI one can visualize and evaluate all myocardial segments in different planes. Clinically used parameters to express regional myocardial function are wall thickness, systolic wall thickening, and circumferential and longitudinal wall shortening. Wall thickening can be quantified by delineating the endocardial and epicardial borders at end diastole and end systole. Visual analysis of wall motion using semiquantitative scoring for regional wall motion is often used clinically. Different grades are used: normokinesis (normal wall motion), hypokinesis (decreased wall motion), akinesis (absent wall motion), dyskinesis (wall motion in the opposite direction of expected), and hyperkinesis (increased wall motion). With cardiac MRI, wall motion by measuring the amount of centripetal motion throughout systole can be quantified.
Myocardial function at rest can be normal; however, during periods of increased oxygen demand, myocardial ischemia resulting in regional myocardial dysfunction, expressed as wall motion abnormalities, may ensue. This can be assessed with gradient-echo cine MRI. Dobutamine stress cine MRI has been reported to be a useful test in adults with coronary artery disease, particularly in patients with poor acoustic windows. More recently, the use of an MRI-compatible supine cycle ergometer has been reported in patients with congenital heart disease to allow assessment of ventricular function and valve regurgitation response to exercise. Stress MRI also allows evaluation of contractile reserve. Contractile reserve is the magnitude of augmentation of ventricular performance with stress. The study of contractile reserve with stress MRI demonstrates functional abnormalities that remain occult at rest and thereby helps to detect early ventricular dysfunction. This may be particularly useful in the evaluation of RV function after correction of tetralogy of Fallot or after the atrial switch procedure. Stress MRI may also detect the functional consequences of coronary anomalies.
Cardiac MRI examinations combine a comprehensive description of anatomy together with the quantitative evaluation ventricular function in patients with heart disease. The evaluation of ventricular volumes and function in patients after repair of tetralogy of Fallot is the most common referral for cardiac MRI examinations in pediatric patients. An accurate assessment of ventricular volumes and function is particularly helpful in these patients because timely detection and monitoring of changes in RV volume have prognostic implications. It has been shown that there is a lower likelihood of postoperative reduction in RV end-diastolic volume if pulmonary valve replacement is performed when preoperative RV end-diastolic volume is larger than 170 mL/m2. Also, cardiac MRI can provide detailed delineation of RV outflow tract anatomy, which is important for planning surgical or interventional valve replacement. In addition, velocity-encoded phase-contrast MRI can be used to quantify pulmonary regurgitation in these patients.
Quantifying systemic RV volumes and systolic function in patients after atrial switch operations or patients with congenitally corrected TGA can be challenging because these ventricles are dilated and hypertrophied with multiple prominent trabeculations. MRI is the only technique that allows acquisition of full-volume data sets encompassing the entire subaortic right ventricle. In addition, cardiac MRI can be used to demonstrate the 3D anatomy of the venous baffles and assess flow.
Patients with arrhythmogenic RV cardiomyopathy develop RV dysfunction and are at increased risk of sudden death. Although often challenging and subjective, cardiac MRI can detect subtle wall motion abnormalities, which occur in the early phase of the disease. Also, fatty fibrous tissue infiltration of the RV free wall can also be demonstrated with a T1 fast-spin echo sequence.
Because of its robustness, cardiac MRI is especially well suited for follow-up of patients with valvular heart disease in which an accurate evaluation of ventricular volumes and ejection fraction is crucial for surgical timing.
In patients with simple congenital heart lesions such as atrial septal defects, ventricular septal defects, and patent ductus arteriosus, cardiac MRI can be used to quantify Qp/Qs and to calculate ventricular volume as an indicator of the hemodynamic burden imposed by the defect.
Other situations in which cardiac MRI examinations are performed in the assessment of ventricular function include patients with univentricular hearts with limited acoustic windows. Cardiac MRI allows evaluation of ventricular volumes and function, assessment of the Fontan baffle status, patency of a fenestration, presence of thrombi, and measurement of pulmonary blood flow. It can also be optimal to detect pulmonary venous obstruction, evaluate atrio-ventricular valve function, and determine the presence of outflow tract obstruction.
Detailed preexamination MRI planning is crucial, given the wide array of imaging sequences available and the complex nature of the clinical, anatomic, and functional issues in patients with congenital heart disease. As with echocardiography, cardiac MRI examination of patients with congenital heart disease is an interactive diagnostic procedure that requires online review and interpretation of the data by the supervising physician. The imaging specialist need not be deterred by the anatomic variability found in these patients. Discussions with a pediatric cardiologist can clarify the clinical questions to be answered, enabling a comprehensive anatomic and diagnostic examination.
To aid in reproducibility and accelerate clinical scanning, MRI assessment of ventricular function is performed in a highly standardized manner, producing two long-axis cine images and a stack of 10 to 12 short-axis cines (see Fig. 4.30). A cine image can be acquired in one breath hold of about 8 to 10 s. Therefore, a typical set of ventricular images can be acquired in less than 5 minutes. Initially, a series of transverse pilot views are obtained. From these, the mitral valve and LV apex can be identified. The vertical long-axis (VLA) cine image can then be obtained. The VLA is then used to plan the horizontal long-axis (HLA) image, which shows all four chambers and the mitral and tricuspid valves. The HLA and VLA views are then used to plan the short axis. The first short-axis plane should be placed using the end-diastolic image at the base of the heart covering the most basal portion of the left and right ventricles just forward of the atrioventricular ring. It is vital that the basal slice is acquired in a consistent and standardized manner to optimize reproducibility. Further short-axis images are acquired sequentially every 6 to 8 mm in children and every 8 to 10 mm in adults along the long axis to the cardiac apex. All cines should be acquired during breath holding. The expiratory phase of respiration has shown to provide more reproducible diaphragmatic positions, but acquiring the images in inspiration is better tolerated.
RV “short-axis” volume data are available as a byproduct of the LV volume short-axis acquisition. However, this is not true of RV short-axis data. For true RV short-axis data, the stack of short-axis slices should be specifically planned so that they are parallel to the tricuspid valve. As a result, extra care is needed when defining and analyzing the most basal slice of the above conventional RV “short-axis” data. The RV volume measurements can alternatively be made from data sets acquired in the axial orientation (Video 4.15). We acquire these data from the coronal localizing images, which demonstrate the gross cardiac anatomy, by planning an orthogonal stack of slices to cover the heart from a level just below the diaphragm to the pulmonary bifurcation (Fig. 4.31). It has been demonstrated that RV volume calculations are slightly more accurate when using axial images.
Additionally, velocity-encoded cine MRI can be acquired in the transverse plane crossing the proximal ascending aorta or proximal pulmonary artery. This makes it possible to quantify the hemodynamic severity of shunts and calculate stroke volume in a way independent of the volumetric calculation, and thus can be used as an internal control of the accuracy of the measurements. Care should be taken to set the limit of the velocity encoding range above the expected values to avoid aliasing. The velocities within the myocardium have recently been measured in 3D using velocity-encoded cine MRI by setting low velocity encoding ranges (15 to 30 cm/s). This method, however, lacks sufficient temporal resolution and has blood-related artifacts. These problems may be resolved by future improvements in image acquisition techniques.
Cardiac MRI image analysis is usually performed off-line using commercially available analysis software. This remains a time-consuming process requiring manual delimitation of multiple images. For LV and RV volume analyses, end diastole and end systole are visually defined as the phase with the largest area (generally the first) and the phase with the smallest LV and RV areas, respectively. The endocardial contour has to be drawn in each slice at end diastole and end systole. Also, the epicardial boundary has to be delimited (generally in end diastole) to calculate LV mass. Ideally, the papillary muscles should be outlined separately and excluded from the ventricular volume but included in the myocardial mass. This, however, is not always possible. At the base of the heart, slices are considered to be within the left ventricle if the blood volume is surrounded by 50% or more ventricular myocardium. If the basal slice contains both ventricular and atrial tissue, the ventricular contours are drawn to the junction with the atrium and joined by a straight line through the blood pool. The workstation computes the LV and RV end-diastolic volume, end-systolic volume, stroke volume, ejection fraction, and LV mass using a modified Simpson’s rule equation.
Figure 4.31. Right ventricular axial scan in a patient after arterial switch operation for transposition of the great arteries. Plane scans extend from the diaphragm (top left) to above the pulmonary artery (bottom right).
The manual delineation of LV endocardial and epicardial borders takes 10 to 15 minutes per data set. Development of algorithms for automatic border detection has facilitated the application of these techniques, but further refinements are required to improve their efficiency.
Through-plane motion represents a major challenge to achieve accurate volume measurements during the cardiac cycle. Through-plane motion results from a 15% to 18% longitudinal shortening in normal hearts while the base of the heart moves toward the apex. As a consequence of this long-axis shortening, a short-axis slice positioned through the base of the left ventricle at end diastole will be located in the atrium at end systole. Without correction for through-plane motion, end-systolic volumes are overestimated and ejection fraction is underestimated. Although correction for through-plane motion can be achieved with slice tracking techniques, this promising feature is not clinically available. So correction techniques need to be applied during postprocessing.
Problems and Potential Solutions
Despite many advantages, the use of cardiac MRI for the evaluation of cardiac function in pediatric patients with heart disease is still limited by several factors. MRI cannot be performed in patients with recent implantation of vascular coils (within 6 weeks). Pacemakers and defibrillators are also contraindicated. Sternal wires, vascular coils, prosthetic heart valves, and intracardiac devices will induce imaging artifacts, mainly when using gradient-echo sequences.
Cardiac MRI evaluation under general anesthesia or deep sedation is needed in patients younger than 7 to 8 years or in subjects with claustrophobia. Other patients may be unable to hold their breath for long periods (longer than 10 s). Performing cardiac MRI with free breathing can be performed in patients who are unable to hold their breath using two approaches. The first is to use a diaphragmatic navigator so that an image is only acquired when the diaphragm, and therefore the heart, is in a preset location. More recently, “real-time” imaging has been used. In this case, images are acquired in one cardiac cycle, rather than across several. Real-time images are of lower resolution but still able to render reproducible information.
Irregular heart rhythms such as atrial fibrillation or frequent ectopy pose a challenge to the MRI acquisition because images are acquired over several heartbeats. This requires acquiring the images at comparable RR intervals. Blurred images result from acquiring images when the RR intervals are irregular. The use of real-time imaging can overcome these problems.
Future applications of MRI in the setting of pediatric heart disease include optimization of current sequences to make acquisition faster while maintaining spatial resolution. More interactive planning with more “online” information and shorter postprocessing time would be desirable as well.
In addition, the use of new sequences may provide additional valuable information. Postcontrast myocardial enhancement has been extensively validated and proved to be helpful in defining viable myocardium in patients with ischemic heart disease and demonstrating the presence of myocardial scar in various conditions such as dilated and hypertrophic cardiomyopathies. The value of this technique to detect areas of myocardial fibrosis in patients with congenital heart disease warrants further evaluation. Also, evaluation of contractile reserve and myocardial perfusion with exercise may help to identify early cardiac dysfunction and areas of abnormal perfusion, respectively. Finally, high field-strength magnets (3T) with multiple channels and faster gradients will allow better imaging of small children and smaller structures like the coronary arteries.
Interventional MRI is a new, exciting field. The combination of 3D information with high spatial resolution and real-time imaging makes MRI particularly appealing as a guide to intravascular procedures such as (re)coarctation or pulmonary valve stenting. MRI-guided cardiac catheterization allows simultaneous acquisition of MRI flow data and invasive pressure measurements and thus accurate measurements of pulmonary vascular resistance.
Significant recent technological developments allow the clinician to use a wide array of different methods and techniques to assess systolic ventricular function in children. No technique, however, allows the easy quantification of intrinsic myocardial contractility. Most techniques used in daily clinical practice are load-dependent and/or geometry-dependent. An individualized approach considering the clinical question to be answered and the benefits and drawbacks of each technique appears appropriate.
Bhave NM, Lang RM. Evaluation of left ventricular structure and function by three-dimensional echocardiography. Current Opinion in Critical Care 2013;19:387–396.
Blessberger H, Binder T. Non-invasive imaging: Two dimensional speckle tracking echocardiography: basic principles. Heart 2010;96:716–722.
Cheung YF, Hong WJ, Chan GC, Wong SJ, Ha SY. Left ventricular myocardial deformation and mechanical dyssynchrony in children with normal ventricular shortening fraction after anthracycline therapy. Heart 2010;96:1137–1141.
Didier D, Ratib O, Beghetti M, et al. Morphologic and functional evaluation of congenital heart disease by magnetic resonance imaging. J Magn Reson Imaging 1999;10:639–655.
Dragulescu A, Grosse-Wortmann L, Fackoury C, et al. Echocardiographic assessment of right ventricular volumes after surgical repair of tetralogy of Fallot: clinical validation of a new echocardiographic method. J Am Soc Echocardiogr 2011;24:1191–1198.
Fogel MA, Weinberg PM, Chin AJ, Fellows KE, Hoffman EA. Late ventricular geometry and performance changes of functional single ventricle throughout staged Fontan reconstruction assessed by magnetic resonance imaging. J Am Coll Cardiol 1996; 28:212–221.
Fogel MA, Weinberg PM, Gupta KB, et al. Mechanics of the single left ventricle: a study in ventricular-ventricular interaction II. Circulation 1998;98:330–338.
Fogel MA. Assessment of cardiac function by magnetic resonance imaging. Pediatr Cardiol 2000;21:59–69.
Forsey J, Friedberg MK, Mertens L. Speckle tracking echocardiography in pediatric and congenital heart disease. Echocardiography 2013;30:447–459.
Friedberg MK, Mertens L. Echocardiographic assessment of ventricular synchrony in congenital and acquired heart disease in children. Echocardiography 2013;30:460–471.
Geyer H, Caracciolo G, Abe H, et al. Assessment of myocardial mechanics using speckle tracking echocardiography: fundamentals and clinical applications. J Am Soc Echocardiogr 2010;23:351–69; quiz 453–455.
Grothues F, Smith GC, Moon JC, et al. Comparison of interstudy reproducibility of cardiovascular magnetic resonance with two-dimensional echocardiography in normal subjects and in patients with heart failure or left ventricular hypertrophy. Am J Cardiol 2002;90:29–34.
Jasaityte R, Heyde B, D’Hooge J. Current state of three-dimensional myocardial strain estimation using echocardiography. J Am Soc Echocardiogr 2013;26:15–28.
Jenkins C, Chan J, Bricknell K, Strudwick M, Marwick TH. Reproducibility of right ventricular volumes and ejection fraction using real-time three-dimensional echocardiography: comparison with cardiac MRI. Chest 2007;131:1844–1851.
Jenkins C, Moir S, Chan J, Rakhit D, Haluska B, Marwick TH. Left ventricular volume measurement with echocardiography: a comparison of left ventricular opacification, three-dimensional echocardiography, or both with magnetic resonance imaging. Eur Heart J 2009;30:98–106.
Kaku K, Takeuchi M, Tsang W, et al. Age-related normal range of left ventricular strain and torsion using three-dimensional speckle-tracking echocardiography. J Am Soc Echocardiogr 2014;27:55–64.
Kapetanakis S, Bhan A, Murgatroyd F, et al. Real-time 3D echo in patient selection for cardiac resynchronization therapy. JACC Cardiovasc Imaging 2011;4:16–26.
Kutty S, Graney BA, Khoo NS, et al. Serial assessment of right ventricular volume and function in surgically palliated hypoplastic left heart syndrome using real-time transthoracic three-dimensional echocardiography. J Am Soc Echocardiogr 2012;25:682–689.
Lang RM, Badano LP, Tsang W, et al. EAE/ASE recommendations for image acquisition and display using three-dimensional echocardiography. J Am Soc Echocardiogr. 2012;25:3–46.
Lopez L, Colan SD, Frommelt PC, et al. Recommendations for quantification methods during the performance of a pediatric echocardiogram: a report from the Pediatric Measurements Writing Group of the American Society of Echocardiography Pediatric and Congenital Heart Disease Council. J Am Soc Echocardiogr 2010;23:465–495.
Lorenz CH. The range of normal values of cardiovascular structures in infants, children, and adolescents measured by magnetic resonance imaging. Pediatr Cardiol 2000;21:37–46.
Lu X, Nadvoretskiy V, Bu L, et al. Accuracy and reproducibility of real-time three-dimensional echocardiography for assessment of right ventricular volumes and ejection fraction in children. J Am Soc Echocardiogr 2008;21:84–89.
Marcus KA, Mavinkurve-Groothuis AM, Barends M, et al. Reference values for myocardial two-dimensional strain echocardiography in a healthy pediatric and young adult cohort. J Am Soc Echocardiogr 2011;24:625–636.
Mertens LL, Friedberg MK. Imaging the right ventricle—current state of the art. Nature Reviews Cardiology 2010;7:551–563.
Monaghan MJ. Role of real time 3D echocardiography in evaluating the left ventricle. Heart. 2006;92:131–136.
Mor-Avi V, Jenkins C, Kuhl HP, et al. Real-time 3-dimensional echocardiographic quantification of left ventricular volumes: multicenter study for validation with magnetic resonance imaging and investigation of sources of error. JACC Cardiovasc Imaging 2008;1:413–423.
Mor-Avi V, Lang RM, Badano LP, et al. Current and Evolving Echocardiographic Techniques for the Quantitative Evaluation of Cardiac Mechanics: ASE/EAE Consensus Statement on Methodology and Indications Endorsed by the Japanese Society of Echocardiography. Eur J Echocardiogr 2011;12: 167–205.
Nagueh SF, Appleton CP, Gillebert TC, et al. Recommendations for the evaluation of left ventricular diastolic function by echocardiography. J Am Soc Echocardiogr 2009;22:107–133.
Niemann PS, Pinho L, Balbach T, et al. Anatomically oriented right ventricular volume measurements with dynamic three-dimensional echocardiography validated by 3-Tesla magnetic resonance imaging. J Am Coll Cardiol 2007;50:1668–1676.
Notomi Y, Srinath G, Shiota T, et al. Maturational and adaptive modulation of left ventricular torsional biomechanics: Doppler tissue imaging observation from infancy to adulthood. Circulation 2006;113:2534–2541.
Oosterhof T, van Straten A, Vliegen HW, et al. Preoperative thresholds for pulmonary valve replacement in patients with corrected tetralogy of Fallot using cardiovascular magnetic resonance. Circulation 2007;116:545–551.
Pettersen E, Helle-Valle T, Edvardsen T, et al. Contraction pattern of the systemic right ventricle shift from longitudinal to circumferential shortening and absent global ventricular torsion. J Am Coll Cardiol 2007;49:2450–2456.
Sengupta PP, Tajik AJ, Chandrasekaran K, Khandheria BK. Twist mechanics of the left ventricle: principles and application. JACC Cardiovascular Imaging 2008;1:366–376.
van der Zwaan HB, Helbing WA, McGhie JS, et al. Clinical value of real-time three-dimensional echocardiography for right ventricular quantification in congenital heart disease: validation with cardiac magnetic resonance imaging. J Am Soc Echocardiogr 2010;23:134–140.
Vogel M, Schmidt MR, Kristiansen SB, et al. Validation of myocardial acceleration during isovolumic contraction as a novel noninvasive index of right ventricular contractility: comparison with ventricular pressure-volume relations in an animal model. Circulation 2002;105:1693–1699.
1.Which of the following is an advantage of 3D echocardiography in the assessment of ventricular function?
A.Higher temporal resolution
B.Higher spatial resolution
C.Avoids foreshortening of the left ventricular apex
D.Absence of artifacts
E.Use of smaller probes
2.Which of the following can be calculated from an echocardiographic 3-D dataset?
A.Left-ventricular end-diastolic volume
B.Left-ventricular end-systolic volume
C.Left-ventricular ejection fraction
E.All of the above
3.Which part of the right ventricle is most difficult to image and may be excluded from a 3-D echocardiography dataset?
C.The tricuspid valve
D.The outflow tract
E.The pulmonic valve
4.Which of the following is an advantage of cardiovascular MRI?
A.Ability to image the entire heart
B.Higher spatial resolution than CT
C.Can be performed in every patient
D.MRI is a bedside technique
E.Analysis is automated
5.Myocardial tagging is performed to calculate which of the following?
A.Left-ventricular end-diastolic volume
B.Left-ventricular end-systolic volume
C.Left-ventricular ejection fraction
E.- Regional myocardial deformation
6.Which of the following can NOT be calculated from a velocity-encoded, phase-contrast MRI sequence acquired across the proximal main pulmonary artery?
A.Right-ventricular stroke volume
B.Right-ventricular end-systolic volume
C.Pulmonary valve regurgitant fraction
D.Pulmonary valve regurgitant volume
E.Peak gradient across the pulmonary valve
7.The traces below show which of the following?
A.Globally reduced longitudinal myocardial deformation
B.Normal global radial myocardial deformation
C.Normal global longitudinal myocardial deformation
D.Decreased circumferential myocardial deformation
E.Inferior myocardial infarction
8.Which of the following is shown in the myocardial deformation traces shown below?
A.Reduced shortening and presence of post-systolic shortening in the mid interventricular septum
B.Globally reduced longitudinal strain
C.Reduced longitudinal strain in the inferior wall due to a right coronary artery infarction
D.Dyssynchrony due to left bundle branch block
9.The septal flash can be seen in which of the following?
A.Previous inferior myocardial infarction
B.Systemic right ventricle
C.Right ventricular volume overload
D.Presence of dyssynchrony due to left bundle branch block
10.Which of the following is a main advantage of speckle tracking-derived myocardial deformation over tissue Doppler-derived myocardial deformation analysis?
A.It is more sensitive to detect subtle changes.
B.It has higher temporal resolution.
C.Analysis has been standardized for all vendors.
D.It has higher spatial resolution, therefore can detect changes in small regions.
E.It is independent from angle of insonation.
1.Answer: C. Three-dimensional imaging allows the operator to guide the imaging plane through the LV annulus to the LV apex; therefore, foreshortening of the left-ventricular apex can be avoided. Temporal and spatial resolutions of 3-D imaging are lower than that of 2-D imaging. Three-dimensional imaging is prone to stitching artifacts and requires the use of bigger probes than 2-D imaging.
2.Answer: E. All of the above can be calculated from a 3-D echocardiographic dataset.
3.Answer: D. Because of the fact that the width of the 3-D pyramidal datasets needs to be restricted to maintain temporal and spatial resolution, not infrequently the right ventricular outflow tract, which is the most anterior structure of the RV, cannot be included as part of the RV datasets.
4.Answer: A. Cardiovascular MRI is able to image the entire heart from any imaging window. Cardiovascular MRI has lower spatial resolution than CT. It cannot be performed bedside. Several patients have contraindications to undergoing cardiovascular MRI and its analysis requires delineation or at least correction of endocardial and epicardial borders by an operator.
5.Answer: E. Myocardial deformation in the radial, circumferential and longitudinal directions can be calculated when analyzing tagging images. The other parameters are calculated with steady-state free precession images.
6.Answer: B. One can calculate RV stroke volume, pulmonary regurgitant fraction, and volume as well as peak velocity and gradient across the pulmonary valve from a velocity-encoded phase contrast MRI sequence. However, the calculation of RV end-systolic volume requires imaging of the entire RV using steady-state free precession sequence.
7.Answer: C. A normal longitudinal myocardial deformation pattern from a four-chamber view is shown. The average peak systolic longitudinal strain is more negative than -18%. The values are slightly higher at the LV apex than the base. Finally, a homogeneous pattern with all myocardial segments peaking around the time of aortic valve closure is seen.
8.Answer: A. Longitudinal myocardial deformation traces at the basal, mid, and apical interventricular septum are shown. The mid segment (yellow trace) shows severely reduced systolic deformation with post-systolic shortening occurring after aortic valve closure. The other segments shorten normally. This is commonly seen in the most hypertrophied segments in patients with hypertrophic cardiomyopathy.
9.Answer: D. Activation of the interventricular septum at the beginning of systole is seen in patients with left bundle branch block and dyssynchrony. This phenomenon is called “septal flash.” Inferior myocardial infarctions do not generally lead to left bundle branch block. A systemic right ventricle and right ventricular volume overload are associated with right bundle branch block which does not cause early systolic septal activation.
10.Answer: E. Speckle tracking-derived analysis of myocardial deformation is independent from the angle of insonation. It has not been proven to be more sensitive nor has higher temporal or spatial resolution than tissue Doppler-derived myocardial deformation analysis. Analysis packages have not been standardized for all vendors.