Author + information
- Received November 28, 2011
- Accepted December 13, 2011
- Published online March 1, 2012.
- Paaladinesh Thavendiranathan, MD, MSc⁎,
- Shizhen Liu, MD, PhD⁎,
- David Verhaert, MD⁎,
- Anna Calleja, MD⁎,
- Adrien Nitinunu⁎,
- Thomas Van Houten, RDCS⁎,
- Nathalie De Michelis, BSN†,
- Orlando Simonetti, PhD⁎,
- Sanjay Rajagopalan, MD⁎,
- Thomas Ryan, MD⁎ and
- Mani A. Vannan, MBBS⁎,⁎ ()
- ↵⁎Reprint requests and correspondence to:
Dr. Mani A. Vannan, Professor of Clinical Medicine, The Ohio State University, 473 West 12th Avenue, DHLRI, Suite 200, Columbus, Ohio 43210
This work was presented at the 2011 American College of Cardiology Young Investigators competition.
Objectives To assess the feasibility, accuracy, and reproducibility of real-time full-volume 3-dimensional transthoracic echocardiography (3D RT-VTTE) to measure left ventricular (LV) volumes and ejection fraction (EF) using a fully automated endocardial contouring algorithm and to identify and automatically correct the contours to obtain accurate LV volumes in sinus rhythm and atrial fibrillation (AF).
Background 3D transthoracic echocardiography is not used routinely to quantify LV volumes and EF. A fully automated workflow using RT-VTTE may improve clinical adoption.
Methods RT-VTTE was performed and 3D EF and volumes obtained using an automated trabecular endocardial contouring algorithm; an automated correction was applied to track the compacted myocardium. Cardiac magnetic resonance (CMR) and 2-dimensional biplane Simpson method were the reference standard.
Results Ninety-one patients (67 in normal sinus rhythm [NSR], 24 in AF) were included. Among all NSR patients, there was excellent correlation between RT-VTTE and CMR for end-diastolic volume (EDV), end-systolic volume (ESV), and EF (r = 0.90, 0.96, and 0.98, respectively; p < 0.001). In patients with EF ≥50% (n = 36), EDV and ESV were underestimated by 10.7 ± 17.5 ml (p = 0.001) and by 4.1 ± 6.1 ml (p < 0.001), respectively. In those with EF <50% (n = 31), EDV and ESV were underestimated by 25.7 ± 32.7 ml (p < 0.001) and by 16.2 ± 24.0 ml (p = 0.001). Automated contour correction to track the compacted myocardium eliminated mean volume differences between RT-VTTE and CMR. In patients with AF, LV volumes and EF were accurate by RT-VTTE (r = 0.94, 0.94, and 0.91 for EDV, ESV, and EF, respectively; p < 0.001). Automated 3D LV volumes and EF were highly reproducible.
Conclusions Rapid, accurate, and reproducible EF can be obtained by RT-VTTE in NSR and AF patients by using an automated trabecular edge contouring algorithm. Furthermore, automated contour correction to detect the compacted myocardium yields accurate and reproducible 3D LV volumes.
- atrial fibrillation
- automated contour correction
- automated contour detection
- cardiac magnetic resonance
- real-time full volume 3D volume transthoracic echocardiography
Three-dimensional (3D) transthoracic echocardiography (TTE) is more accurate and reproducible than 2-dimensional (2D) TTE for the evaluation of left ventricular (LV) volumes and ejection fraction (EF) (1,2). However, limitations to the adoption of 3D TTE in clinical practice include the need to stitch together gated subvolumes and the need to semiautomatically contour the endocardial borders for volumes and EF, which is time-consuming and reduces reproducibility (3). In addition, several studies have illustrated the underestimation of LV volumes by 3D TTE when compared with cardiac magnetic resonance (CMR) (4,5). Although the potential reasons for this underestimation have been addressed (4,5), it is unclear as to which factor plays a predominant role. Also, accurate assessment of LV volumes and EF in patients with atrial fibrillation (AF) is important (6), but these patients have been excluded in previous 3D TTE studies because of the need for gated acquisition.
Real-time full-volume 3-dimensional transthoracic echocardiography (RT-VTTE) is a new imaging technique by which the left ventricle can be imaged continuously (every cardiac cycle, nonstitched) in its entirety using a 90° × 90° volume sector (Online Video 1). Fully automated contouring of the 3D LV endocardial surface can be done by using a knowledge-based probabilistic contouring algorithm (7).
The goals of this study were to: 1) assess the feasibility, accuracy, and reproducibility of RT-VTTE to measure LV volumes and EF using a fully automated endocardial contouring algorithm compared with CMR in patients with normal sinus rhythm (NSR); 2) identify and automatically correct the most important source of discrepancy in LV volumes between CMR and RT-VTTE; and 3) assess the feasibility and accuracy of RT-VTTE for multiple heartbeat, automated, LV volumes and EF measurements in patients with AF compared with the 2D biplane Simpson method.
For study goals 1 and 2, patients in NSR who were >18 years of age referred for a clinically indicated CMR were enrolled. Inclusion criteria were as follows: 1) CMR study with segmented breath-held cine images; and (2) good acoustic windows for RT-VTTE imaging (determined before any analysis), defined as the ability to image the entire left ventricle in any apical orientation without dropouts or artifacts. For study goal 3, patients >18 years of age in AF referred for a 2D TTE were enrolled. For test–retest reproducibility, a separate group of patients referred for a 2D TTE in NSR were enrolled. The study was approved by an institutional review board at the Ohio State University.
CMR imaging and analysis
CMR was performed using a 1.5-T magnet (MAGNETOM Avanto; Siemens Medical Solutions, Erlangen, Germany) using a 12-channel phased-array coil. Short-axis cines extending from the mitral valve plane to just below the LV apex (8) were acquired using a segmented balanced steady-state free precession sequence, effective temporal resolution of 20 to 25 frames per cardiac cycle, and 8-mm short-axis slices with 20% interslice distance.
Images were analyzed using Argus software (Siemens Medical Solutions), following current guidelines (8), by observers blinded to RT-VTTE data. Short-axis slices with at least 50% of the LV circumference surrounded by myocardium were included in the volume (9). The basal and apical slices were confirmed on long-axis views. The LV cavity was traced both in end-diastole (ED) and end-systole (ES) with inclusion of the papillary muscle and trabeculae in the volume (10,11). End-diastolic volume (EDV) and end-systolic volume (ESV) as well as EF were calculated using the Simpson method.
RT-VTTE acquisition and quantification
RT-VTTE was performed using the Acuson SC2000 (Siemens Ultrasound, Mountain View, California) imaging system with a 4Z1c real-time volume transducer (2.8 MHz). Full LV volumes were acquired at every cardiac cycle for 3 to 5 consecutive cycles. Care was taken to include the entire left ventricle within the imaging volume using automatically generated simultaneous 2D reference planes (Online Video 2).
Analysis of the RT-VTTE images was performed off-line using a fully automated knowledge-based endocardial detection algorithm (7,12) with prototype software (eSie LVA, Siemens Ultrasound) by a blinded investigator. This algorithm automatically detects the endocardial surface from knowledge gained from large, expert-annotated training databases of volume data combined with a 3D discriminative model to match relevant image features of the given LV volume to the database. EDV was selected by the algorithm using the peak R-wave from the electrocardiography signal and ESV as the systolic frame with the minimal volume. The LV cavity (including the papillary muscles) was then displayed as a 3D mesh rendering and on 2D planes for visualization (Fig. 1, Online Videos 3 and 4). Although the ED/ES frames and endocardial contours could be modified, no manual corrections were made for our primary analysis. Time volume curves were automatically generated for all the cardiac cycles, and EDV, ESV, and EF are displayed. The cardiac cycle with the best endocardial detection was chosen for analysis. This was defined as: 1) all myocardial segments (including the apex) were well tracked; 2) the contours approximate the visually assessed endocardial edge; and 3) the mitral valve plane was optimally tracked.
Accuracy of the Automated Contouring Algorithm for Measurement of LV Volumes and EF
To first determine the accuracy of the automated LV contouring algorithm, it was applied to CMR short-axis cines reconstructed in a volume format identical to RT-VTTE datasets from 35 randomly selected patients (Online Video 5). The LV volumes and EF derived by the algorithm were compared with the manual CMR method (reference standard).
Accuracy of Automated LV Volume and EF Measured With RT-VTTE Compared With CMR in NSR
Having verified the accuracy of the automated contouring algorithm, we applied it to RT-VTTE data to calculate LV volumes and EF and compared it with CMR values for all patients together and then separately in patients with normal and reduced EF (13). The ability to correctly classify patients with EF <50% and <35% was also assessed compared with CMR. These cutoff points were chosen because they identify patients with reduced EF (<50%) (13) and those who qualify for device therapy in heart failure (EF <35%).
Sources for Discrepancy in LV Volumes Between RT-VTTE and the Manual CMR Method
In CMR, the contours are prescribed at the compacted myocardium, including the trabeculae in the ventricular volumes (Fig. 2A) (11). In echocardiography, the contours are often drawn at the trabeculae–blood pool interface or within the trabeculae (Fig. 2B) (10). To investigate the impact of this difference in techniques, a 0.5-, 1.0-, and 1.5-mm automated contour correction was applied to the RT-VTTE trabecular contours generated by the algorithm (Fig. 2C), which moves the contours outward toward the compacted myocardium and recalculates volumes and EF.
Another source of difference in volumes is the inclusion/exclusion of the LV basal short-axis slices in CMR. Because we included basal slices in which at least 50% of the circumference is surrounded by myocardial tissue, we would include some volume that encompasses the mitral annulus (9) (Fig. 3A). To avoid this, some centers only include slices below the mitral annulus (9), similar to RT-VTTE (4). To determine how much this contributes to discrepancies in volume, 2 investigators (in consensus and blinded to CMR volumes) manually moved the basal contours in RT-VTTE images to include the annular plane (Fig. 3B [red lines]) in ED and ES, as appropriate, in 20 randomly selected patients. The volumes and EF were recalculated and compared with CMR.
Reproducibility of RT-VTTE and CMR Volumes and EF for Patients in NSR
Two investigators independently reviewed the 3 to 5 cardiac cycles for each patient and recorded EDV, ESV, and EF from the best cycle. Volumes and EF were then compared. One of the investigators repeated this process 3 months later for intraobserver variability.
For a selected cardiac cycle, when automated contour correction was applied, there was no significant difference in the final volume or EF between 2 observers. To compare this finding with the reproducibility of manual contour corrections, 2 investigators independently modified the automated trabecular contours in 35 randomly selected patients to track the compacted myocardium (interobserver variability) for a selected cardiac cycle as noted here. This test was repeated by an investigator 1 month later (intraobserver variability). In addition, CMR interobserver and intraobserver variability was measured in 35 patients.
In 22 separate patients, test–retest reproducibility was assessed by first obtaining a full volume of the left ventricle, followed by repositioning the patient and the transducer and obtaining a second full volume by a different sonographer. EDV, ESV, and EF were measured in a fully automated manner as described here, and the data from the last cardiac cycle in every patient were used for analysis.
Accuracy of RT-VTTE in AF
RT-VTTE was analyzed for each heartbeat in 3 to 5 consecutive cardiac cycles (Fig. 4). The analysis method was identical to that described for patients with NSR. 2D apical 2- and 4-chamber planes generated from the volume dataset corresponding to the same cardiac cycles were independently analyzed using the biplane Simpson method for volume and EF. The data from these 2 methods were compared at a beat-to-beat level and at the patient level (averaged over the acquired cardiac cycles).
The RT-VTTE and CMR volumes and EF were compared using linear regression and Pearson correlation coefficient. Bland-Altman analysis with paired t tests was used to assess the bias and limits of agreement with CMR. Interobserver and intraobserver variability was calculated as the absolute difference of the corresponding pair of repeated measurements as a percentage of their mean in each patient and then averaged over the study group. Test–retest reproducibility analysis consisted of Pearson r, Bland-Altman analysis, and paired t tests. The paired t test was also used to compare ventricular volumes and EF between the 2 modalities. Unpaired t tests and the chi-square test were used when appropriate.
Patient demographics and feasibility of RT-VTTE
A total of 145 consecutive patients were screened for inclusion in the main study; 10 (7%) refused to participate and 37 (26%) were excluded due to poor echocardiography window or artifacts and 7 (5%) for incomplete acquisition due to technical errors. The analysis algorithm worked successfully in all 91 included patients (67 had NSR and 24 had AF). RT-VTTE was completed within 1 h of the CMR study. Demographic characteristics of the included patients are shown in Table 1. Acquisition of 3 to 5 consecutive cardiac cycles of RT-VTTE datasets in any one apical orientation required 5 to 10 s and was heart rate dependent. The mean temporal resolution of the RT-VTTE images was 45 ± 27 ms (32 ± 20 volumes/s [VPS]), with only 2 patients with volume rates <10 VPS (8 and 9 VPS). Analysis of ventricular volumes and EF took 30 to 60 s for 3 to 5 cardiac cycles.
Accuracy of the automated LV contouring algorithm with reconstructed 3D CMR data
The EDV, ESV, and EF obtained using the automated algorithm with the reconstructed 3D CMR datasets were accurate compared with the manual CMR method (reference standard) (Fig. 5). The correlation coefficients for EDV, ESV, and EF were 0.99, 0.99, and 0.95, respectively, with biases of –2.4 ± 9.6 ml (p = 0 .15), 0.6 ± 7.4 ml (p = 0.61), and –0.8 ± 4.7% (p = 0.35).
Accuracy of the automated LV contouring algorithm with RT-VTTE in NSR
Among all patients in NSR, RT-VTTE and CMR measurements correlated highly for EDV, ESV, and EF (r = 0.90, 0.96, and 0.98, respectively; p < 0.001 for all) (Fig. 6, Table 2). EDV and ESV were underestimated by 17.6 ± 26.7 ml and 9.8 ± 17.9 ml (p < 0.001), respectively; however, EF was accurate (0.3 ± 2.5%; p = 0.27). In the subgroup of patients with EF ≥50%, the RT-VTTE and CMR measurements also correlated highly for EDV, ESV, and EF (r = 0.85, 0.92, and 0.86, respectively; p < 0.001 for all) (Table 3). However, EDV and ESV were underestimated by 10.7 ± 17.5 ml (p = 0.001) and by 4.1 ± 6.1 ml (p < 0.001), whereas EF was accurate (–0.5 ± 2.5%; p = 0.3). In the subgroup of patients with EF <50%, the respective r values were 0.88, 0.93, and 0.97 (p < 0.001 for all) (Table 3). EDV and ESV were underestimated by 25.7 ± 32.7 ml (p < 0.001) and by 16.2 ± 24.0 ml (p = 0.001), whereas EF was accurate (–0.2 ± 2.4%; p = 0.7). With respect to classifying patients with an EF cutoff of <50% and <35%, RT-VTTE accurately identified 64 of 67 patients (96%) and 65 of 67 patients (97%), respectively.
Accuracy of RT-VTTE volumes after automated contour correction in NSR
Automated contour correction applied to the original trabecular contour to detect the compacted myocardium resulted in improved accuracy of EDV and ESV by RT-VTTE (Table 3). In patients with EF ≥50%, 1.0-mm and 0.5-mm automated contour correction and in patients with EF <50%, 1.5-mm correction to EDV and ESV, respectively, provided the best agreement with CMR. With these automated corrections, there were no statistically significant differences in the mean volumes between RT-VTTE and CMR.
Accuracy of RT-VTTE volumes after basal plane adjustment in NSR
Inclusion of the mitral annular plane in RT-VTTE volumes (Fig. 3) decreased the volume underestimation by 7 ml and 4 ml for EDV and ESV, respectively. Although the EDV and ESV biases were reduced, the absolute difference between RT-VTTE and CMR remained significant (Table 4).
Reproducibility of volumes and EF measurements
Among the 67 patients in NSR, the mean interobserver variability for EDV, ESV, and EF were 1.0 ± 4.3%, 0.8 ± 4.0%, and 0.4 ± 4.5%, respectively. The intraobserver variability was 0.4 ± 1.9%, 0.2 ± 0.9%, and 0.3 ± 1.3%.
In the 35 patients in whom manual corrections were made, the mean underestimation in RV-VTTE EDV and ESV decreased from 13 ± 17% to 8 ± 16% and from 13 ± 17% to 6 ± 17%, respectively. However, the volumes were still lower than the CMR volumes (p < 0.05). EF was not statistically different before or after contour correction. The mean SD interobserver and intraobserver variability for the manually corrected EDV, ESV, and EF are summarized in Table 5. The variability was higher at the individual patient level, and the CMR reproducibility was good.
The mean age of the included patients for test–retest reproducibility analysis was 48 ± 19 years; 59% were male; and mean volume rate of the RT-VTTE acquisitions was 37 ± 21 VPS (range 12 to 74 VPS). The mean EDV, ESV, and EF for the first and second measurements were as follows: 140 ± 46 ml versus 139 ± 45 ml (p = 0.37), 77 ± 44 ml versus 77 ± 43 ml (p = 0.93), and 47 ± 15% versus 47 ± 15% (p = 0.56). The respective Pearson r values were 0.99, 0.99, and 0.98. Bland-Altman analysis showed minimal bias between the 2 measurements and good levels of agreement (Fig. 7).
Accuracy of RT-VTTE in AF
In 24 of 27 (89%) patients with AF, RT-VTTE images were of diagnostic quality. At a beat-to-beat level, automated trabecular contours by RT-VTTE and manual contours by 2D biplane Simpson methods correlated highly for EDV, ESV, and EF (r = 0.93, 0.93, and 0.92, respectively; p < 0.001 for all). At the patient level, the corresponding correlation coefficients were 0.94, 0.94, and 0.91 (p < 0.001 for all) (Fig. 8). The mean EDV, ESV, and EF by RT-VTTE were 123 ± 46 ml, 71 ± 40 ml, and 44 ± 10%, whereas by 2D biplane Simpson method, they were 122 ± 45 ml, 66 ± 33 ml, and 46 ± 9% (p > 0.05 for all). On Bland-Altman analysis, at the patient level, the biases and levels of agreement for EDV, ESV, and EF were 2 ± 16 ml, 4 ± 13 ml, and –2 ± 4% (p > 0.05 for all) (Fig. 8). The mean heart rate during image acquisition was 82 ± 19 beats/min.
This is the first clinical study to assess the feasibility, accuracy, and reproducibility of RT-VTTE to measure EF and volumes using a fully automated contouring algorithm in patients with NSR and AF. We demonstrated that: 1) RT-VTTE coupled with a fully automated endocardial contouring algorithm to measure LV volumes and EF is feasible, accurate, and reproducible; 2) the use of trabecular edge versus compacted myocardium for LV volumes is the most important reason for volume discrepancy between RT-VTTE and CMR; 3) an automated contour correction algorithm applied to RT-VTTE to track the compacted myocardium improves the accuracy of LV volumes compared with CMR and is reproducible; and 4) it is now possible to measure 3D LV volumes and EF in patients with AF accurately and rapidly. The implementation of such an automated workflow for quantitative 3D TTE of the LV may promote wider use in routine clinical practice.
The average temporal resolution of RT-VTTE was 45 ± 27 ms, with a mean temporal resolution of 24 ms with harmonic imaging and 14 ms with fundamental imaging in excellent acoustic windows. This range of “real-time” temporal resolution is both acceptable and adequate for imaging the left ventricle (8). We were able to use data from 71% (91 of 128) of the imaged patients for analysis of LV function. The exclusion rate is a reflection of patients being recruited from the CMR laboratory who may have been initially referred due to poor acoustic windows. We were able to use data from 89% of the patients with AF who were recruited from the TTE laboratory, consistent with previously reported feasibility rates in quantitative 2D TTE (14). Lastly, our exclusion rate is not different from previous 3D studies (5,15,16).
Although image quality is important for clinical adoption, quantitative 3D TTE is not performed in current practice even when there is excellent image quality due to lack of intuitive workflow. We were able to obtain accurate 3D LV EF and volumes for multiple cardiac cycles in about 30 to 60 s in an entirely automated manner compared with 4 to 10 min previously reported with manual/semiautomated approaches with gated 3D (17,18). This workflow is a significant step forward in facilitating wider clinical adoption of 3D TTE for measurement of LV function.
Automated EF measurements with RT-VTTE were accurate both in patients with normal and reduced EF compared with use of CMR. Our EF estimates were better than that reported in a recent study using single-beat real-time 3D TTE, with significantly lower volume rates and manual contour corrections (16). The LV volumes in our study with normal and reduced EF were underestimated compared with CMR using automated trabecular contouring. However, this is consistent with previous single-center, gated 3D studies (5,15,19) and significantly less than the 67 ml and 41 ml underestimations seen in the only multicenter study of gated 3D TTE (4). Also, none of these previous studies reported separate values for patients with normal and reduced EF. We analyzed patients with reduced EF separately because these patients have abnormal LV anatomy, and it is necessary to assess the accuracy of the automated algorithm specifically in these patients. In addition, the volume underestimation in our study seemed to be higher in patients with severely dilated ventricles (20,21) (Fig. 6), consistent with findings in other 3D studies (4). This finding may be a reflection of the small sample size of patients with severe ventricular dilation in our study or the need for improvement in the algorithm with severely dilated ventricles.
To improve the accuracy of volume measurements, we applied an automated contour correction to track the compacted myocardium instead of the trabecular edge. In patients with reduced EF, the best LV volume estimates by RT-VTTE compared with CMR were obtained by 1.5-mm correction of EDV and ESV contours, whereas 1.0-mm and 0.5-mm corrections were necessary for EDV and ESV contours, respectively, in patients with normal EF. The need to correct the contours in both ED and ES in patients with reduced EF may lie in the fact that the distinction between the trabecular endocardium and the compacted myocardium is exaggerated in the dilated dysfunctional left ventricle in both ED and ES (Fig. 9 [illustrated with CMR images]). The minimal contour correction that was necessary in normal EF is not clinically significant and is likely not necessary in daily practice, as supported by the minimal underestimation in EDV and ESV compared with CMR in our study (10 ml and 4 ml). However, contour correction in patients with reduced EF will likely be necessary until improvements in endocardial tracking are made and/or development of an algorithm occurs that can track compacted myocardial contours in contrast 3D TTE. However, the latter may not be possible in all patients because there are contraindications to contrast use. Alternatively, manual corrections to the contours could be made at the cost of reduced reproducibility and tedious workflow, which are significant impediments to routine clinical adoption of 3D TTE. With respect to other sources of discrepancy between volumes obtained by RT-VTTE and CMR, the inclusion of mitral annular plane in CMR was not as important as differences in contouring techniques (Tables 1 and 3).
In previous gated 3D TTE studies, the interobserver variability has ranged from 0.1% to 8.2% for EDV, 2.1% to 13.5% for ESV, and 0% to 13.1% for EF (4,5,22). Even at expert centers, the variation in LV volumes at the patient level was as high as 38% to 70% (4). In our study with automated workflow, the interobserver and intraobserver variability for volumes and EF was ≤1%. Manual contour corrections resulted in larger interobserver and intraobserver variability and did not resolve the significant underestimation of volumes, suggesting that expert manual contouring may not be ideal for clinical application. In addition, the automated technique had excellent test–retest reproducibility, a feature that has seldom been reported in previous 3D studies.
In patients with AF, LV volumes and EF derived by automated contouring of RT-VTTE and 2D biplane Simpson method showed excellent correlation for EDV, ESV, and EF at a beat-to-beat level and averaged over heartbeats at the patient level. The 2D planes from the 3D volumes were chosen as the reference standard because it is not possible to obtain segmented balanced steady-state free precession CMR cine images in patients with AF for volumetric analysis. Contrary to previously published reports (5) showing systematic underestimation of LV volumes by 2D biplane Simpson method compared with gated 3D TTE, a good agreement between the 2D method and RT-VTTE was found in our study. This finding may be because the 2D images were obtained from the same 3D dataset, hence minimizing any differences in acquisition technique or discrepant cardiac cycle durations. RT-VTTE with automated contouring affords a rapid way to measure 3D LV function beat-by-beat in patients with AF.
We only included patients with good acoustic windows, which results in selection bias. However, given that our study is the first to clinically validate the benefit of RT-VTTE using a fully automated contouring algorithm, we felt that this approach was appropriate. Furthermore, our data provide insight into the real-world challenges of suboptimal acoustic windows, which is a physical reality for both 2D and 3D TTE. Secondly, the uniform contour correction algorithm used may not account for regional variations in LV trabecular thickness (Fig. 9), although we did include patients with regional abnormalities, and the automated algorithm and correction yielded accurate LV volumes and EF. There is perhaps a minority in whom profound variations in regional geometry may adversely affect this automated algorithm. We also did not compare the automated 3D algorithm with an automated 2D algorithm that has been described previously (23,24). However, 3D imaging is superior to 2D methods for LV function and volumes (2), especially in patients with abnormal LV dilation or distorted shapes. In addition, although the automated 2D algorithm eliminates the need for manual contouring of the endocardial border, it does not account for the fundamental limitations of 2D echocardiography, which includes the need for assumption of ventricular shape and apical foreshortening. In fact, studies using automated 2D algorithms have shown poor correlation with CMR volumes (24), with significant volume underestimation despite manual contour adjustments (23). Finally, the variability in the choice of the basal slice and the need for manual contouring for volumes and EF calculations is a limitation of CMR.
RT-VTTE is a significant improvement over gated 3D TTE because it overcomes many technical and practical limitations of gated 3D TTE. Accurate and reproducible EF can be obtained by RT-VTTE in patients with NSR and AF using an automated trabecular edge contouring algorithm. Furthermore, appropriate application of an automated contour correction algorithm to detect the compacted myocardium yields accurate and reproducible 3D LV volumes. Such an automated workflow, which is also time efficient, may enhance the adoption of quantitative 3D TTE of the left ventricle in routine clinical echocardiography when the acoustic window is optimal.
We acknowledge Bogdan Georgescu, PhD, Helene Houle, RDCS, FASE, and Joel Mancina, RDCS, from Siemens who played an important role in the development of the automated contouring algorithm and training of the clinical sonographers. We also thank Nicholas A. Tomson, RDCS, RVT, for help with clinical RT-VTTE image acquisition.
For supplementary videos and their legends, please see the online version of this article.
Dr. Simonetti has received research grants from Siemens Healthcare. Dr. Vannan is a member of the advisory board and receives a per diem, a speaker's honorarium, and research support from Siemens. D.r. Ryan is on the advisory board for Philips. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- cardiac magnetic resonance
- end-diastolic volume
- ejection fraction
- end-systolic volume
- left ventricular
- normal sinus rhythm
- real-time full-volume 3-dimensional transthoracic echocardiography
- transthoracic echocardiography
- Received November 28, 2011.
- Accepted December 13, 2011.
- American College of Cardiology Foundation
- Mor-Avi V.,
- Jenkins C.,
- Kuhl H.P.,
- et al.
- Jenkins C.,
- Moir S.,
- Chan J.,
- Rakhit D.,
- Haluska B.,
- Marwick T.H.
- Miyasaka Y.,
- Barnes M.E.,
- Gersh B.J.,
- et al.
- Yang L.,
- Georgescu B.,
- Zheng Y.,
- Meer P.,
- Comaniciu D.
- Macron L.,
- Lim P.,
- Bensaid A.,
- et al.
- Soliman O.I.,
- Krenning B.J.,
- Geleijnse M.L.,
- et al.
- Kuhl H.P.,
- Schreckenberg M.,
- Rulands D.,
- et al.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.
- Cannesson M.,
- Tanabe M.,
- Suffoletto M.S.,
- et al.