Quantitative analysis of segmental myocardial deformation of different myocardial layers has become possible using strain-encoded cardiac magnetic resonance imaging (SENC) and speckle-tracking echocardiography (STE). We evaluated and compared the quantitative analysis of myocardial deformation using SENC and STE.
In 44 patients (age 61 ± 13 years, 34 men), SENC by cardiac magnetic resonance imaging using a 1.5-Tesla whole-body scanner and two-dimensional STE were performed prospectively. Quantitative layer-specific analysis of segmental left ventricular function was performed to determine the peak circumferential and peak longitudinal systolic strain values using SENC and STE of an endocardial and epicardial myocardial layer. In addition, segmental function was defined as normokinetic, hypokinetic, or akinetic by visual analysis of the magnetic resonance imaging cine sequences.
The endocardial and epicardial strain defined by SENC or STE differed significantly between the visually defined segmental function states. The correlation of the peak circumferential endocardial strain by SENC versus STE (intraclass correlation coefficient [ICC] 0.493, 95% CI 0.358–0.597) tended to be better than the correlation of the circumferential epicardial strain using both methods (ICC 0.321, 95% CI 0.238–0.399). The correlation of the peak longitudinal endocardial strain by SENC and STE was similar (ICC 0.472, 95% CI 0.398–0.541), in contrast to the longitudinal epicardial strain analysis by both techniques (ICC 0.554, 95% CI 0.417–0.655). Circumferential strain analysis by STE allowed better distinction of the hypokinetic or akinetic segments from the normokinetic segments than did the circumferential strain analysis by SENC of the endocardial layer (area under the receiver operating characteristic curve [AUC ROC] 0.946 vs 0.884; P < .001) or epicardial layer (AUC ROC 0.884 vs 0.782; P < .001). Longitudinal strain analysis using STE and SENC of the endocardial layer (AUC ROC 0.851 vs 0.839; P = .5838) and epicardial layer (AUC ROC 0.849 vs 0.833; P = .4321) had similar diagnostic value for identifying the presence of hypokinetic and akinetic segments.
Quantitative analysis of segmental deformation by SENC and STE allowed accurate distinction of myocardial segments with different functional states. Circumferential endocardial strain analysis by STE allowed the best distinction of segments with impaired function from the normokinetic segments.
Quantification of global and regional left ventricular function is an important objective for any cardiac imaging technique. Strain analysis enables the objective quantification of segmental myocardial deformation. Cardiac magnetic resonance imaging (MRI) and echocardiography have been described as imaging modalities allowing strain analysis. For definition of myocardial deformation using MRI, tagging has been described as the reference method. However, this method requires long breath-hold acquisition times and time-consuming postprocessing. Recently, strain-encoded cardiac magnetic resonance imaging (SENC) has been introduced as a novel MRI technique to measure the myocardial deformation, expressed as the myocardial circumferential and longitudinal strain using harmonic phase MRI. This is in contrast to MRI tagging, which uses special pulse sequences of saturated magnetization that are traditionally oriented orthogonally to the image plane. In MRI tagging, detailed motion of the myocardium can be deduced by analyzing the deformation of the tag lines found within these images. SENC was found to be comparable to MRI tagging for the definition of regional function. Furthermore, SENC has been reported to define myocardial infarct transmurality. It has been shown to be superior to MRI tagging in terms of temporal resolution, total scan duration, and the time required for postprocessing of the acquired data during dobutamine stress.
Improvements in two-dimensional (2D) echocardiographic image resolution have enabled detection and tracking of acoustic markers from frame to frame, allowing analysis of the radial, circumferential, and longitudinal strain. Most imaging techniques consider the complete myocardial wall thickness in the analysis of myocardial function without additional distinction between the different layers of the myocardium. In particular, the effect of ischemic cardiomyopathy can be expressed differently in the endocardial and epicardial myocardial layers, because the endocardial layers are affected first by flow-limiting coronary stenosis. Thus, a separate analysis of myocardial function for the different myocardial layers is of interest. A separate analysis of myocardial deformation for the endocardial and epicardial layer has been described using SENC. Recently, layer-specific myocardial deformation has also become available using 2D speckle-tracking echocardiography (STE). We sought to evaluate and compare SENC and STE for the analysis of endocardial and epicardial layer-specific myocardial deformation.
A total of 49 patients (age 61 ± 13 years, 34 men) were prospectively included in the present study. They underwent cardiac MRI to determine left ventricular function, evaluate inducible myocardial ischemia, or assess myocardial viability. In all patients, echocardiography was performed within 24 hours after MRI. The SENC and 2D STE analyses were performed on the acquired images. Of the 49 patients, 5 (10.2%) had to be excluded because of insufficient echocardiographic image quality for subsequent quantitative analysis. The remaining 44 patients formed the study group. The patient characteristics and indications for MRI are listed in Table 1 .
|Age (years)||61 ± 13|
|End-diastolic volume (mL)||199 ± 100|
|End-systolic volume (mL)||127 ± 102|
|Ejection fraction (%)||43 ± 18|
|Previous ST-segment elevation myocardial infarction||7|
|Previous non–ST-segment elevation myocardial infarction||14|
|Coronary artery disease||3|
The MRI studies were performed with a 1.5 Tesla whole-body MRI scanner (Achieva, Philips Medical Systems, Best, The Netherlands) using a 32-channel cardiac coil. Assessment of the left ventricular function at rest was determined from the cine images using steady-state, free-precession sequences in 10–12 slices of 8-mm thickness, covering the whole left ventricle from the base to the apex, and long-axis, 2- and 4-chamber views. The parameters were as follows: echocardiographic time 1.39 ms, repetition time 2.8 ms, flip angle 60°, spatial resolution 2.4 × 2.5 × 8 mm 3 , and temporal resolution 21–28 ms, with a breath-hold time of 7–10 sec/image. The pulse sequence was prospectively gated. Wall motion analysis was performed using the MRI cine sequences. Regional function of the basal, midventricular, and apical segments of the septal, anterior, lateral, and inferior wall was evaluated visually by 2 experienced readers as normokinetic, hypokinetic, or akinetic. In the case of a disagreement between the 2 readers, a third reader was asked to resolve the issue.
Strain-encoded cardiac MRI is a special modification of the MRI scanner software that enables the quantification of regional tissue deformation caused by cardiac motion. To calculate myocardial strain, SENC uses tag planes parallel, not orthogonal, to the image plane. Therefore, 2- and 4-chamber views in the same plane orientation as that of the cine images were generated to calculate the circumferential strain and three short-axis views (basal, midventricular, and apical) were acquired to measure longitudinal strain (typical imaging parameters: voxel size 4/4/10 mm 3 , repetition time/echocardiographic time 25/0.9 ms, and flip angle 30°). The temporal resolution was set at 25 ms. The number of cardiac phases was adapted to 26–38 to cover approximately 85–90% of the cardiac cycle. The total time for scanning the strain data was 10–14 seconds. For SENC, the circumferential strain was calculated from the long-axis views, and the longitudinal strain was calculated from the short-axis views, because SENC uses tag planes parallel, and not orthogonal, to the image plane. All strain measurements were performed using the SENC images and dedicated software (Diagnosoft MAIN, version 1.06, Diagnosoft, Palo Alto, CA). The circumferential and longitudinal strain was calculated for each segment ( Figure 1 ).
Echocardiographic studies were performed with a Vivid Seven System (GE Vingmed, Horton, Norway). Parasternal short-axis views at the basal, midventricular, and apical levels and apical 2- and 4-chamber views were acquired with a frame rate of at least 50 frames/sec. Circumferential strain, as a parameter of circumferential deformation determined by echocardiography, relates to motion along the curvature of the left ventricle in the parasternal short axis, and longitudinal strain relates to motion along the longitudinal axis. Circumferential and longitudinal shortening during systole involves a decrease in the length of a myocardial segment and, thus, is expressed in negative numbers. Hence, an improvement in strain is indicated by a larger negative number. Echocardiographic analysis of the three acquired parasternal short-axis views and two apical views was performed off-line with the aid of a dedicated software package (EchoPAC, GE Vingmed, Haifa, Israel). This system allows analysis of the peak systolic circumferential strain for six segments within one short-axis cut plane and analysis of the peak longitudinal strain of six segments within one long-axis cut plane according to the detection of acoustic speckles, as previously described. The system automatically determines the tracking quality for each analyzed segment and automatically tracks the endocardial and epicardial line on high-quality images. In the present study, segments with suboptimal tracking quality, as defined by the system, were excluded from the analysis ( Figure 2 ).
Quantitative Analysis of Myocardial Deformation
Quantitative layer-specific analysis of segmental left ventricular function was performed according to the definition of peak systolic circumferential and peak systolic longitudinal strain of an endocardial and epicardial myocardial layer using SENC and STE. For this purpose, the system automatically divided the wall thickness into three layers of equal thickness, an endocardial, a midmyocardial, and an epicardial layer. High-spatial resolution has been described for SENC and STE, allowing analysis of myocardial layers of 3-mm thickness.
The analysis was performed on 12 of 16 segments, considering the 16-segment model of the American Society of Echocardiography. These 12 segments included septal basal, midventricular and apical, anterior basal, midventricular and apical, lateral basal, midventricular and apical and inferior basal, and midventricular and apical segments in all 44 patients.
In 15 randomly selected studies, the strain analysis by SENC and STE was repeated by another experienced observer to determine the interobserver agreement.
The statistical analysis was performed with a special statistical analysis program (MedCalc Software, version 18.104.22.168, Mariakerke, Belgium). Continuous data are presented as the mean ± SD and were compared using Student’s t test or analysis of variance as adequate, with post hoc analysis using the Student-Newman-Keuls test for all pairwise comparisons. Pearson’s correlation coefficient ( r ) and intraclass correlation coefficient (ICC) were calculated to express agreement between the strain defined using SENC and that using STE. The endocardial and epicardial strain values determined using SENC were compared with those from STE using linear regression analysis, with presentation of the standard error of estimate (SEE), and Bland-Altman analysis demonstrating bias, with 95% CIs for all segments. The receiver operating curves of peak systolic circumferential strain of the endocardial and epicardial layer determined by SENC and STE were calculated for the distinction of the hypokinetic or akinetic segments from the normokinetic segments. The interobserver agreement was demonstrated by calculating the coefficient of variation of repeated measurements and ICC. P < .05 was considered significant.
The analysis included 528 myocardial segments. For the circumferential strain analysis, 34 segments (6.4%) had to be excluded owing to suboptimal parasternal short-axis echocardiographic tracking quality. For the longitudinal strain analysis, 63 segments (11.9%) had to be excluded because of suboptimal apical echocardiographic tracking quality. No segment had to be excluded because of insufficient image quality by SENC.
Visual Wall Motion Analysis of MRI Cine Sequences
Of the 528 myocardial segments, 311 (58.9%) of the septal, anterior, lateral, and inferior walls were classified as normokinetic by visual analysis of the wall motion, considering the MRI cine sequences obtained in the 2- and 4-chamber views. Of the 217 remaining, 132 (25.0%) were classified as hypokinetic and 85 (16.1%) as akinetic.
Comparison of SENC and STE
The correlation of the peak circumferential endocardial strain using SENC compared with STE ( r = 0.64, SEE 4.8%, ICC 0.493, 95% CI 0.358–0.597) tended to be better than the correlation of the circumferential epicardial strain by both methods ( r = 0.36, SEE 4.5%, ICC 0.321, 95% CI 0.238–0.399). The correlation of the peak longitudinal endocardial strain by SENC and STE was similar ( r = 0.53, SEE 8.2%, ICC 0.472, 95% CI 0.398–0.541) compared with the longitudinal epicardial strain analysis using both techniques ( r = 0.61, SEE 5.5%, ICC 0.554, 95% CI 0.417–0.655; Figure 3 ).
The peak circumferential endocardial strain using STE was greater than the circumferential endocardial strain using SENC (bias 3.7%, 95% CI −13.8% to 21.2%, P < .001). The circumferential epicardial strain analyzed using the two methods was also significantly different. However, the difference was less (bias between STE and SENC, −1.5%, 95% CI −15.1% to 12.2%; P < .001).
The peak longitudinal strain by STE was greater than the longitudinal strain by SENC in the endocardial layer (bias between STE and SENC −0.8%, −17.1% to 15.4%; P = .0294) and in the epicardial layer (bias between STE and SENC −2.4%; −13.6% to 8.8%; P < .001).
Differentiation between Segments with Different Function States
The endocardial and epicardial peak systolic circumferential strain determined by SENC and STE was significantly different among the normokinetic, hypokinetic, and akinetic segments defined by visual wall motion analysis of the MRI cine sequences ( Table 2 ). Similar differences were found in the analysis of longitudinal strain by SENC and STE ( Table 3 and Figure 4 ). Considering the normokinetic segments, the peak circumferential strain was greater using STE than with SENC in the endocardial layer (bias between STE and SENC 7.3%, 95% CI −10.0% to 24.7%; P < .0001) and the epicardial layer (bias 0.2%, 95% CI −14.0% to 14.4%; P < .0001). The peak longitudinal strain was not significantly different between STE and SENC, considering the normokinetic segments in the endocardial layer (bias between STE and SENC 0.8%, 95% CI −16.4% to 17.9%; P = .1438) and the peak longitudinal strain lower using STE in the epicardial layer (bias −1.8%, 95% CI −13.3% to 9.7%; P < .0001; Figure 5 ).
|Wall motion by MRI cine sequences||SENC CS||STE CS|
|Endocardial layer||Epicardial layer||Endocardial layer||Epicardial layer|
|Normokinesia (n = 288)||−20.5% ± 4.0%||−15.2% ± 4.1%||−27.8% ± 8.7%||−15.4% ± 6.3%|
|Hypokinesia (n = 125)||−14.1% ± 4.8%||−11.1% ± 4.2%||−12.7 ± 5.8%||−8.0% ± 3.9%|
|Akinesia (n = 81)||−9.2% ± 4.9%||−9.6% ± 4.3%||−7.8% ± 6.1%||−4.5% ± 4.3%|
|ANOVA||< .001||< .001||< .001||< .001|
|Post hoc analysis ∗||< .05||< .05||< .05||< .05|
|Wall motion by MRI cine sequences||SENC LS||STE LS|
|Endocardial layer||Epicardial layer||Endocardial layer||Epicardial layer|
|Normokinesia (n = 285)||−18.8% ± 5.1%||−16.2% ± 5.0%||−19.7% ± 8.6%||−14.9% ± 6.1%|
|Hypokinesia (n = 119)||−12.9% ± 4.5%||−10.8% ± 4.2%||−9.5% ± 6.3%||−7.7% ± 5.2%|
|Akinesia (n = 61)||−10.1% ± 4.6%||−8.2% ± 4.1%||−5.8% ± 6.7%||−4.0% ± 4.2%|
|ANOVA||< .001||< .001||< .001||< .001|
|Post hoc analysis ∗||< .05||< .05||< .05||< .05|