Cardiac imaging has become fully integrated in the management of patients, with echocardiography being the most widely used modality. Two-dimensional echocardiography (2DE) offers a detailed assessment of ventricular function, valvular pathology, and hemodynamics, but its comprehensiveness is limited by the fact that images represent a single plane, and may thus inadequately represent a heart whose function is not uniform. For example, when evaluating 2-D images to calculate ventricular volumes, geometrical assumptions need to be made regarding the ventricular shape, which may not be valid in the presence of regional wall motion abnormalities. More than 15 years ago, 3DE was developed to provide a more accurate assessment of ventricular volume, mass, and function to enhance the ability to assess the spatial relations between cardiac structures and provide a more complete view of the valves. While multiple studies demonstrated the advantages of 3DE, it too had several limitations. 3-D images were actually reconstructed from multiplanar 2-D images taken over several cardiac cycles. This increased the risk of introducing artifact from patient respiration or motion. Analysis was performed off-line and required tedious manual tracing of endocardial borders. In addition, limited computer technology caused slow data processing. As a result, 3DE remained a research tool that was not used widely in the clinical arena.
The past years have seen significant developments in transducer technology, allowing pyramidal shaped blocks to be acquired in realtime. It is now possible to image an entire ventricle or valve and analyze it online, thus obviating the need for complicated and time-consuming reconstructions. Advances in computer technology have also contributed to streamlining and expediting data analysis. Realtime 3-D echocardiography (RT3DE) has thus greatly enhanced the potential clinical utility for the evaluation of left ventricular (LV) and valvular function. This article examines the technological improvements in RT3DE and the evidence to support its routine use in the clinical setting.
Early models of RT3DE transducers used only 256 elements, which did not fire simultaneously, resulting in an image quality that was often inferior to that of 2DE. Displayed images still consisted of computer-generated 2-D cut planes obtained from the 3-D dataset, and the size of the pyramidal scan volume was incapable of accommodating larger ventricles. The recently developed matrix array transducer uses 3,000 simultaneously firing elements, and, as a result, contrast resolution and penetration has improved.
In addition, rendered 3-D images can now be displayed online. The size of the pyramidal volumes and the method in which they are acquired depend on what structure is being imaged (see Figure 1).To capture the entire LV volume, four wedges of 93° x 20° are obtained over four cardiac cycles during a breath hold. Each sub-volume is triggered to the R wave of every other beat to allow sufficient time for the probe to recalibrate and each sub-volume to be stored. The ability to image patients with arrhythmias or dyspnea is therefore limited. However, if the area of interest is a valve, a more narrow pyramid of data can be magnified, without the need for respiratory or cardiac gating.
Evaluation of LV Parameters
Assessment of LV volume, ejection fraction (EF), and mass has become a fundamental component of decision-making regarding the risk for future cardiac events and the timing of interventions such as valvular surgery. While magnetic resonance imaging (MRI) has been the gold standard for determining LV parameters, factors such as the relatively high cost and long examination time with multiple breath-holds make this modality less than ideal for serial examinations. 2DE has traditionally been used as a more feasible alternative, but the ability to obtain accurate values is dependent on the ability to acquire non-foreshortened long-axis images from apical acoustic windows. In many patients, this is compromised due to limited access to the true ventricular apex. Analysis of 2DE can be problematic, because objective measurements are often substituted with 'eye-balled' estimates, which are subjective and experience-dependent. Attempts at quantitative analysis by using manual tracings of 2-D images are also limited, because they rely on geometric assumptions of chamber size and shape, which may not apply to patients with abnormalities such as aneurysms or hypertrophy.
Recent studies using both animal models and human subjects have confirmed that RT3DE can both accurately and reproducibly evaluate LV parameters. RT3DE can image the entire ventricle, without forcing the reader to make assumptions regarding size and shape or subjectively select views for analysis (see Figure 2). As a result, when compared with MRI,RT3DE more closely approximates volume than 2DE, although RT3DE continues to slightly underestimate end-diastolic volumes (EDV) and end-systolic volumes (ESV). Possible explanations for this, albeit small, discrepancy between RT3DE and MRI include the different tracing techniques used (long-axis cross sections with RT3DE versus short-axis slices for MRI) and the fact that apical definition is less reliable with both methods. An earlier study evaluating an in vitro model and patients showed that, even in the presence of ventricular aneurysms, there was an excellent correlation with both actual volumes (in the in vitro model) and MRI (in patients).
As with 2DE, it is possible to use contrast with RT3DE to improve visualization of endocardial borders in patients with poor acoustic windows, particularly at the apex. However, one study showed that when contrast was used with RT3DE and continuous imaging, agreement with MRI measurements actually decreased. This was probably because of increased bubble destruction caused by the high density of scan lines required to generate a 3-D data set. A subsequent study used dual triggering at end-systole and end-diastole to overcome this limitation, resulting in more accurate values for ESV, but it continued to slightly underestimate EDV.
RT3DE can also be applied to the evaluation of ventricular mass among patients with a broad range of cardiac diseases, including coronary artery disease, dilated cardiomyopathy, and aortic abnormalities.RT3DE allows the operator to find the true non-foreshortened 2-D cut planes from which to perform measurements and thus minimize errors, which is particularly advantageous in patients with irregularly shaped hearts. This concept was illustrated in one study of LV mass among a group of patients with various cardiac diseases, in which the LV long-axis dimensions were significantly larger with RT3DE compared with 2DE.
Analysis of LV parameters with RT3DE involves using semi-automated border detection, which significantly reduces the amount of time required to analyze each set of images. Caiani et al. found that analysis took, on average, five minutes per dataset, but this time was extended with suboptimal image quality, while Kuhl et al. reported a time of 12 minutes for RT3DE and 14 minutes for MRI. Importantly, it appears that accuracy is not sacrificed when using the semi-automatic method. Kuhl et al. found no significant differences between EDV, ESV, and EF values obtained by manual tracing and the semi-automated border detection.
Once a full-volume assessment of the ventricle is complete, can be applied that information to an analysis of wall motion.When trying to determine the presence and degree of wall motion abnormalities, 2DE relies primarily on the reader's subjective interpretation. In contrast, recent data has shown that RT3DE allows for a more quantitative approach, incorporating volumetric information and how it changes over time. By using the semi-automated border detection to generate a cast of the endocardium, it is possible to quantify the precise volume of different ventricular segments throughout the cardiac cycle.
2DE has served as an extremely useful tool for the evaluation of both abnormal wall motion and myocardial perfusion. However, there are important shortcomings that could be overcome with RT3DE. With 2DE, several views are obtained from more than one acoustic window to assess the necessary segments, but adjacent areas of regional dysfunction may be overlooked. By imaging the entire LV volume and allowing for visualization in multiple planes, RT3DE could, theoretically, provide a better assessment of at-risk or infarcted myocardium.
The advantages of RT3DE readily apply to stress echocardiography, where it provides not only a potentially more accurate assessment of wall motion, but also allows for rapid acquisition of images. When compared with 2DE, RT3DE has been shown to detect a higher proportion of patients with coronary artery disease. The improved sensitivity of RT3DE may be, in part, due to the fact that, during stress testing, there is little time to capture all of the necessary images before the heart rate returns to baseline, and stress-induced abnormalities disappear. RT3DE has the advantage of faster acquisition, because there is no longer the need to acquire views from several different transducer positions.
RT3DE can be used in two ways during stress testing. The first method is using bi-plane imaging, which allows visualization of two orthogonal planes simultaneously in a split-screen display. The primary image, an apical four-chamber view, is first optimized similarly to that of 2DE, and then the secondary image plane can be adjusted by the bi-plane angle being rotated without the transducer having to be moved. Investigators have shown that the time required to obtain all of the required views is decreased, thus allowing for acquisition closer to peak heart rate. This makes it an ideal technique for exercise stress testing, where the need for rapid acquisition is particularly pronounced. However, whether this technique will improve the sensitivity and specificity of stress testing has yet to be determined.
An alternative method applied to dobutamine stress echocardiography (DSE) involves the acquisition of full-volume data sets at every stage of the examination. The images are then automatically cropped off-line to visualize all segments. The application of RT3DE to stress testing is still limited by several technological factors. For instance, the matrix array transducer causes increased bubble destruction when contrast is used to improve the endocardial delineation, leading to decreased visualization of the apical segments. In addition, the large footprint of the matrix array transducer makes it difficult to fit in the intercostal space of smaller patients. Lastly, when the four wedge-shaped sub-volumes are combined to produce a representation of the entire ventricle, stitch artifacts can result in false-positive interpretations due to the erroneous diagnosis of inhomogeneous contraction and relaxation patterns.
When characterizing the structure of a valve, one would ideally utilize a method that examined the valve directly, and was not affected by other factors, such as hemodynamics. Reliable measurements of valve area and function are crucial, given that they help to determine when to recommend operative repair. Most of the studies comparing 2DE and RT3DE focus on mitral valve pathology. With 2DE, planimetry of the mitral valve is the most direct measurement of the valve orifice; however, planimetry can only be obtained in the parasternal view. Restriction to one view in patients with mitral stenosis can make correct plane orientation difficult, because as the disease progresses, the valve develops a more funnel-like shape. Depending on the geometry of the orifice, even a small deviation from the optimal plane can lead to an overestimation of mitral valve area (MVA) of up to 60%. RT3DE allows the operator to rotate the image online—regardless of the acoustic window in which it was acquired—to choose the ideal plane for measurement of the MVA. In patients with mitral stenosis, RT3DE has been shown to have the best agreement with invasively measured MVA when compared with planimetry and continuous-wave Doppler methods. The ability to evaluate the entire valve apparatus without relying on hemodynamics becomes especially important in patients who have had percutaneous mitral valvuloplasty (PMV). Following valvuloplasty, the mitral orifice often becomes irregular, many patients develop an atrial septal defect, and rapid changes occur in left atrial and ventricular compliance post-repair. It is important to note that almost half of the patients in these studies were in atrial fibrillation. While this does not appear to have affected image quality, it did lengthen the acquisition time.