It is well known that the evaluation of right ventricular (RV) volume and function by two-dimensional echocardiography has long been hampered by the complex three-dimensional (3D) shape of this chamber, which has made its geometric modeling challenging and frequently inaccurate. Today, as real-time 3D echocardiographic (RT3DE) imaging is being embraced by echocardiographers around the world as a clinically useful imaging modality with a proven “track record” in the assessment of valvular pathology and left ventricular function, research and clinical applications focused on the right ventricle have been lagging behind. As a result, our progress in understanding how changes in RV size, shape, and dynamics may contribute to pathophysiology has also been relatively slow. The recognition that RT3DE imaging has the potential to overcome these difficulties has recently caused a resurgence of interest in the assessment of this “neglected” chamber. Similar to the technological advancements in the RT3DE evaluation of the left ventricle, an important milestone for the right ventricle was the development of analysis software that allows the semiautomated detection of the 3D RV endocardial surface, from which RV volume can be calculated throughout the cardiac cycle directly, without the need for geometric modeling. Assuming that this analysis is accurate and reproducible, one may see the possibilities for gaining new insights into RV dynamics that will expand the understanding of a spectrum of cardiovascular pathology and facilitate its implementation into routine clinical practice.
The right ventricle is commonly described in the literature as having a “complex crescent shape.” Although the term “crescent” quite accurately reflects the shape of RV cross-section at the apical through midventricular levels, the real complexity of RV shape can be better appreciated at more basal levels, where the elongated inflow and outflow tracts “hug” the base of the left ventricle in a somewhat lopsided manner. Compared with the thick left ventricular myocardium, which is usually easier to differentiate from the blood in the left ventricular cavity on ultrasound images, the precise boundaries of the thinner RV free wall, with its prominent and variably located trabeculae, are more difficult to visualize on either two-dimensional or 3D echocardiographic images ( Figure 1 ). It is well known that different disease states can lead to very different pathologic changes in RV size and shape. An extreme example of such pathologic changes is chronic, severe pulmonary hypertension, which may result in severe RV enlargement, wherein RV volume may double or even triple. Because the two ventricles share one wall (ie, the right ventricle is effectively “anchored” to the interventricular septum), such extreme changes in volume must be accommodated by dramatic changes in the shape of mostly the free wall but also the septum. The ability to identify these signs objectively and quantify them accurately is an important piece in the “diagnostic puzzle” physicians are faced with.
In this issue of JASE , 4 articles describe the use of this new tool to establish the foundations of this methodology and to pave its way into the clinical arena. The study of Leibundgut at al focuses on RT3DE imaging–derived values of RV volume and ejection fraction measured using the new analysis software in a large group of adult patients with normal cardiac anatomy over a wide range of RV size and function, which were compared with those measured within 24 hours using standard magnetic resonance imaging (MRI) methodology based on disc summation. When combined with the studies by van der Zwaan et al and Grewal et al in this issue and the study by Khoo et al in the November 2009 issue of JASE , all with similar goals and designs but in patients with congenital heart disease, these data constitute a solid layer validating this promising new methodology. These studies demonstrate good agreement between RT3DE and MRI measurements, reflected by intertechnique correlations ≥ 0.8, despite a consistent underestimation by echocardiography compared with MRI reference values. Importantly, RV ejection fraction showed only minimal to no significant intermodality differences.
Careful review of the results of these studies reveals several potential reasons for the underestimation of RT3DE imaging–derived RV volumes compared with MRI: (1) the ability to identify the endocardial boundary properly because of limited spatial resolution, (2) the ability to determine accurately where the right ventricle “starts” and “ends” (ie, to identify the tricuspid valve and pulmonic valve planes correctly on both MRI and RT3DE images), and (3) the less than perfect reproducibility of both techniques, as one cannot expect one technique to agree better with a different technique than it agrees with itself on repeated measurements.
Effects of Inaccurate Endocardial Delineation
In patients with congenital heart disease in the study by Khoo et al, the main culprits for the volume underestimation by RT3DE imaging compared with MRI were poor endocardial boundary definition and gain settings that may affect endocardial identification, despite the fact that only half of the consecutive patients were judged to have adequate image quality to be included in the study. This probably reflects the large extent to which we rely on our ability to extrapolate the ventricular boundary where it cannot be clearly visualized, which can be easily “fooled” by anomalous anatomy in these patients. Grewal et al attributed volume underestimation by the RT3DE technique to difficulties with correctly identifying the boundaries of the enlarged, abnormal, and often aneurysmal RV outflow tract in their patients with tetralogy of Fallot or severe pulmonary regurgitation.
In the study by van der Zwaan et al, the degree of RV volume underestimation compared with MRI was larger than that reported by Khoo et al (34% vs 20%), but the percentage of patients included on the basis of sufficient image quality was considerably higher (81% vs 52%), and as a result, accurate boundary identification was less likely in the majority of patients enrolled in the former study.
The volume underestimation by RT3DE compared with MRI reported by Leibundgut et al was attributed to differences in imaging methodology, without offering much detail, which this study was not designed to investigate. This volume underestimation was confirmed in another recent study, in which the potential sources of these differences were specifically targeted. The authors reported that (1) RV volumes were underestimated less compared with MRI reference values obtained using volumetric analysis rather than the disc summation methodology, in agreement with findings recently reported by Nieman et al, and (2) RV volumes were underestimated considerably less in right ventricle–shaped phantoms that were free of endocardial trabeculae than in human ventricles, in which the identification of RV endocardium is challenging because of the suboptimal spatial resolution of RT3DE imaging. Our previous study showed that in the left ventricle, the exclusion of endocardial trabeculae from the MRI reference volume effectively eliminated the underestimation by RT3DE imaging. Van der Zwaan et al tested this hypothesis in the right ventricle in a subgroup of their patients and found that the exclusion of trabeculae from the blood pool reduced the MRI-derived RV volumes by 19 ± 13 mL, which indeed explained most of the difference between RT3DE and MRI measurements in their patients.
The influence of how one deals with endocardial trabeculae becomes increasingly important in disease states in which they are especially prominent, while echocardiographic imaging may be unable to resolve these structures and depict them instead as part of the myocardium. This is especially true during systole, when trabeculae are compressed together. In this regard, contrast enhancement may prove once again useful and allow more accurate delineation of the RV endocardial boundaries by penetrating the intertrabecular spaces. Accordingly, future studies are necessary to investigate the effects of contrast on RV volume measurements.