Background
The aim of this study was to investigate the effect of out-of-plane motion on discrepancies in strain measurements between two-dimensional (2D) and three-dimensional (3D) echocardiography.
Methods
Two-dimensional and 3D data sets were acquired in 54 patients. Using 2D and 3D speckle-tracking software, global circumferential strain (CS) and longitudinal strain (LS) as well as CS and LS at three left ventricular (LV) levels was measured. The effect of through-plane motion was assessed by mitral annular displacement.
Results
Although a good correlation of global CS was noted between the two methods ( r = 0.80, P < .01), mean values of global CS were significantly higher on 3D compared with 2D echocardiography. Correlations of CS and their mean differences were 0.65 and −4.61 at the basal level, 0.76 and −4.17 at the midventricular level, and 0.60 and −2.23 at the apical level, respectively. Correlation of global CS between the two methods was higher in patients who showed mitral annular displacement < 9.4 mm ( r = 0.81) compared with those with mitral annular displacement ≥ 9.4 mm ( r = 0.61). A good correlation of global LS ( r = 0.89, P < .01) was noted, with no significant bias. Correlations of LS and their mean differences were 0.52 and 1.59 at the basal level, 0.89 and −1.17 at the midventricular level, and 0.73 and 1.46 at the apical level, respectively. Correlation of LS between the two methods was higher in patients who showed LV twist < 12.2° ( r = 0.94) compared with patients with LV twist ≥ 12.2° ( r = 0.68).
Conclusions
Through-plane motion produced discrepancies in CS measurements, especially at the LV basal level. Larger bias of LS at the basal and apical LV levels compared with the midventricular level between the two methods suggests that LV twisting also affects the calculation of 2D LS.
Two-dimensional (2D) speckle-tracking echocardiography (STE) is an established technology to assess left and right heart chamber mechanics. Compared with Doppler tissue imaging, it can quantify left ventricular (LV) myocardial deformation in the circumferential, longitudinal, and radial directions with no angle dependency. However, 2D STE has its intrinsic limitations. The use of foreshortened views may affect the accuracy of the quantification of individual components of myocardial deformation. Moreover, the left ventricle contracts in a complex three-dimensional (3D) direction characterized by twisting motion and descent of the base toward the apex during systole. A fixed 2D cutting plane loses original speckles during frame-by-frame speckle-tracking analysis. These problems are encountered for 2D speckle-tracking analysis in both the 2D short-axis and long-axis views. With the advent of 3D STE, speckles in the myocardium can be tracked in 3D space, which has a theoretical advantage to overcome out-of-plane motion.
Several previous studies have investigated the comparison of global and regional strain values between 2D STE and 3D STE. However, there has been a paucity of data regarding the effect of out-of-plane motion on discrepancies in strain measurements between the two methods. We hypothesized that (1) longitudinal movements of the myocardium toward the apex produce the lowest correlation of circumferential strain (CS) measurements at the basal level and the best correlation at the apical level between the two methods, and (2) LV twisting motion produces the lowest correlation of longitudinal strain (LS) measurements at the apical level and the best correlation at the midventricular level because of the helical nature of the myocardium.
Accordingly, we aimed to investigate the effect of longitudinal movements on global and regional CS measurements and the effect of twisting motion on global and regional LS measurements assessed by both 2D STE and 3D STE.
Methods
Study Subjects
From July 1, 2011, to July 31, 2011, we prospectively acquired both 2D and 3D echocardiographic data in patients with a variety of different indications for transthoracic echocardiography. The exclusion criterion was poor image quality that precluded 3D data acquisition from the apical approach. A total of 54 patients were finally entered into the study. Cardiovascular risk factors, including hypertension, ischemic heart disease, atrial fibrillation, diabetes mellitus, hyperlipidemia, and prior myocardial infarction, were evaluated for all patients on the basis of established criteria. The study was approved by the ethics committee in the hospital, and written informed consent was obtained from all patients at the time of echocardiography.
2D Echocardiography
Two-dimensional echocardiographic examinations were performed with subjects in the left lateral decubitus position using a commercially available ultrasound machine and transducer (M5S transducer, Vivid E9; GE Healthcare, Horten, Norway). Three short-axis views at the basal, midventricular, and apical level were acquired for three consecutive beats, with specific attention paid to ensuring that the imaging plane was as circular as possible. The basal short-axis view was located between the tip of the mitral valve and the chordae tendineae. The midventricular short-axis view contained papillary muscle. For the acquisition of the apical short-axis view, the transducer position was shifted to the apex to obtain circular endocardial border without visualization of papillary muscle. Apical four-chamber, two-chamber, and long-axis views to ensure that the long axis of the left ventricle was maximally delineated were also acquired at three consecutive beats. All image acquisitions were performed during breath-holds.
3D Echocardiography
A fully sampled matrix-array transducer (V4 transducer, Vivid E9; GE Healthcare) was used to acquire the 3D data sets. Consecutive six-beat electrocardiographically gated subvolume acquisition was performed from the apical approach using second-harmonic imaging, during apnea to generate the full-volume data set. Care was taken to ensure that the entire LV myocardium was encompassed within the pyramidal data set with as narrow a scan angle as possible. The quality of the acquisitions was verified in each patient by selecting the multislice display mode available on the machines to ensure optimal imaging of the entire LV wall without obvious stitching artifacts. All data sets were digitally stored in a raw-data format and exported to workstations equipped with commercially available quantitative software.
2D Strain Measurements
Using commercially available 2D strain software (EchoPAC PC version 112; GE Healthcare, Milwaukee, WI), the endocardial border in the end-systolic frame was manually traced. A region of interest was then drawn to include the entire myocardium. The software algorithm automatically divided the left ventricle into six equidistant segments and selected suitable speckles in the myocardium for tracking. The software algorithm then tracked the speckle patterns on a frame-by-frame basis using the sum of absolute difference algorithm. Finally, the software automatically generated time-domain LV strain profiles for each of the six segments of each view, from which end-systolic strain was measured. The average value of strain at each level (basal, midventricular, and apical) was calculated. Global CS was determined by averaging the strain values of 18 LV segments. The bull’s-eye diagram from the software package was used to derive global LS with 17 LV segments. If there are more than two uninterpretable segments in a single apical view, the software does not yield global LS.
3D Strain Measurements
Three-dimensional full-volume data sets were analyzed using the 4DAutoLVQ package (EchoPAC PC version 112). The end-diastolic frame displayed in a quad view was used for manual alignment of axis and mitral valve leaflet. Subsequently, in each of the three apical views, three points were indicated manually, including the mitral annular corners at both sides and one apical point. The software automatically traced on the endocardial borders. Manual adjustment was required when software endocardial border detection was suboptimal. LV volumes and ejection fraction were then automatically calculated. Next, to calculate LV mass and myocardial strain, the epicardial border was determined for manual adjustment of the region of interest. The computer then analyzed 3D speckle-tracking in frame-by-frame analysis, with a final 17-segment bull’s-eye map of strain values displayed. In the presence of more than three (of 17) uninterpretable segments, global strain values were not derived.
Strain Analysis
CS
Using 2D and 3D speckle-tracking software, segmental peak systolic CS was measured. Global CS and average CS at each of three LV short-axis levels were calculated and compared in both modalities. Using an anatomic M-mode technique, mitral annular displacement (MAD) at the septal corner of the mitral annulus on the apical four-chamber view was measured, and patients were divided into two groups according to the median value of MAD for investigating the effect of through-plane motion on CS measurements.
LS
Using 2D and 3D speckle-tracking software, segmental peak systolic LS was measured. Global and average LS at three levels of the LV (basal, midventricular, and apical) were calculated and compared between the two methods. In 44 of 54 patients in whom LV twist could be measured from LV short-axis images, patients were divided into two groups according to the median value of LV twist to investigate the effect of LV twisting motion on LS measurements.
Reproducibility of Strain Measurements
To explore the impact of the intrinsic variability of repeated measurements on strain differences, we assessed intraobserver and interobserver variability for 20 randomly selected LV data sets. To test intraobserver variability, a single observer analyzed the same data sets on two different occasions separated by 1 week. To test interobserver variability, a second observer analyzed the data without knowledge of the first observer’s measurements.
Statistical Analysis
Continuous data are expressed as mean ± SD, frequencies, and ranges, as appropriate. Categorical data are presented as numbers or percentages. All statistical analysis was carried out using commercial software (JMP version 9.0, SAS Institute Inc., Cary, NC; SPSS version 17, SPSS, Inc., Chicago, IL). Categorical variables were compared using Fisher’s exact tests or χ 2 tests as appropriate. Student’s t tests were used to test the differences in continuous variables between the two groups. Linear regression analysis was used to study the relationship between two parameters. Bland-Altman analysis was performed to determine bias and limits of agreement between two measurements. Observer variability was assessed using Bland-Altman statistics and intraclass correlation coefficients. P values <.05 were considered statistically significant.
Results
Clinical characteristics of the 54 study patients are given in Table 1 . The mean frame rates of 2D short-axis views and apical views were 61 ± 5 frames/sec (range, 53–68 frames/sec) and 58 ± 4 frames/sec (range, 49–68 frames/sec), respectively. The mean volume rate of 3D full-volume data sets was 33 ± 6 volumes/sec (range, 20–48 volumes/sec).
| Variable | Value |
|---|---|
| Age (y) | 64 ± 18 (15–90) |
| Men/women | 29/25 |
| Heart rate (beats/min) | 67 ± 12 (46–100) |
| Systolic BP (mm Hg) | 141 ± 24 (90–196) |
| Diastolic BP (mm Hg) | 77 ± 13 (55–124) |
| Body surface area (m 2 ) | 1.54 ± 23 (1.13–1.98) |
| Risk factors | |
| Hypertension | 28 (51.9%) |
| Atrial fibrillation | 5 (9.3%) |
| Diabetes mellitus | 13 (24.1%) |
| Hyperlipidemia | 12 (22.2%) |
| Smoking | 17 (31.5%) |
| Clinical diagnosis | |
| Hypertensive heart disease | 8 (14.8%) |
| Ischemic heart disease | 8 (14.8%) |
| Cardiomyopathy | 5 (9.3%) |
| Arrhythmia | 4 (7.4%) |
| Valvular heart disease | 3 (5.6%) |
| Pulmonary hypertension | 2 (3.7%) |
| Heart failure without etiology | 2 (3.7%) |
| Others | 22 (40.7%) |
| Echocardiographic parameters | |
| LVEDVI (mL/m 2 ) | 71 ± 29 (33–180) |
| LVESVI (mL/m 2 ) | 36 ± 27 (10–155) |
| LVEF (%) | 53 ± 13 (14–81) |
Comparison of CS
Adequate image quality of 2D short-axis images for speckle-tracking could not be obtained for at least one level in 10 patients, who were excluded from CS analysis. For regional CS measurements, 2D STE provided adequate tracking in 918 of 972 segments (94.4%), in contrast to 858 of 918 segments (93.4%) for 3D STE ( P < .01). A representative case showing 2D and 3D speckle-tracking echocardiographic measurements of regional CS is shown in Figure 1 . Among 44 patients in whom CS could be measured by both 2D STE and 3D STE, global CS showed a significant correlation between the two methods ( r = 0.80, P < .01; Figure 2 ). However, mean values of global CS were significantly higher on 3D STE compared with 2D STE (−18.4 ± 6.3 vs −14.7 ± 5.0, P < .001). For level basis, correlations of averaged CS and their mean differences were 0.66 and −4.61 at the basal level, 0.76 and −4.17 at the middle level, and 0.60 and −2.23 at the apical level, respectively ( Supplementary Figure 1 [available at www.onlinejase.com ] and Table 2 ). The median value of MAD was 9.4 mm. When patients were divided into two groups according to this cutoff value, the correlation of global and level CS between the two methods was higher in patients with MAD < 9.4 mm compared with those with MAD ≥ 9.4 mm ( Figure 2 , Supplementary Figure 2 [available at www.onlinejase.com ]).
| Parameter | 2D STE | 3D STE | Bias (3D − 2D) | LOA | P |
|---|---|---|---|---|---|
| CS ( n = 44) | |||||
| Global | −14.7 ± 5.0 | −18.4 ± 6.3 | −3.69 | −10.9 to 3.5 | <.0001 |
| Basal | −12.1 ± 5.2 | −16.7 ± 5.6 | −4.61 | −13.4 to 4.1 | <.0001 |
| Midventricular | −15.0 ± 5.7 | −19.1 ± 6.3 | −4.17 | −12.3 to 3.9 | <.0001 |
| Apical | −17.1 ± 7.0 | −19.3 ± 8.7 | −2.23 | −16.7 to 11.9 | .0458 |
| LS ( n = 54) | |||||
| Global | −15.9 ± 5.2 | −15.4 ± 4.6 | 0.42 | −4.1 to 4.4 | .1960 |
| Basal | −14.5 ± 5.7 | −12.9 ± 5.1 | 1.59 | −8.7 to 11.9 | .0317 |
| Midventricular | −15.3 ± 5.3 | −16.5 ± 5.5 | −1.17 | −3.8 to 6.2 | .0016 |
| Apical | −18.4 ± 6.6 | −16.8 ± 6.6 | 1.46 | −7.7 to 10.6 | .0260 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
