Use of myocardial strain to assess global left ventricular function: A comparison with cardiac magnetic resonance and 3-dimensional echocardiography
Article Outline
Background
Ejection fraction (EF) plays a prominent role in clinical decision making but remains dependent on image quality and left ventricular geometry. Using magnetic resonance imaging (MRI-EF) as the reference standard, we sought whether global longitudinal strain (GLS) could be an alternative to the measurement of EF.
Methods
Manual and semi-automated tracing was used to measure Simpson's biplane ejection-fraction (2D-EF) and 3D ejection fraction (3D-EF) and MRI in 62 patients with previous infarction. Global longitudinal strain was measured by 2-dimensional strain (2DS) in the apical views. Automated EF was calculated using speckle tracking to detect the end-diastolic and end-systolic endocardial border.
Results
Strain curves were derived in all segments, with artifactual curves being excluded. The correlation of GLS with MRI-EF (r = −0.69, P < .0001) was comparable to that between 3D-EF and MRI (r = 0.80, P < .0001), and better than that between 2D-EF (r = 0.58, P < .0001) or automated EF and MRI (r = 0.62, P < .0001). To convert GLS into an equivalent MRI-EF, linear regression was used to develop the formula EF = −4.35 ⁎ (strain + 3.9). Of the 32 patients with a normal MRI-EF (≥50%), 75% had normal systolic function by GLS, whereas 85% of patients were recognized as having a normal 3D-EF. Fewer patients were recognized as normal by 2D-EF (70%, P = .14) and automated-EF (61%, P = .04). In those with >6 abnormal segments, the correlation of GLS with MRI-EF improved significantly (r = −0.77, P < .0001) and was similar to 3D-EF (r = 0.76, P < .0001).
Conclusion
Global longitudinal strain is an effective method for quantifying global left ventricular function, particularly in patients with extensive wall motion abnormalities.
Assessment of left ventricular (LV) function is the most frequent indication for echocardiography,1, 2 and ejection fraction (EF) remains the simplest and most widely used parameter for the global assessment of LV function. Ejection fraction is a strong predictor of mortality and retains a central role in clinical decision making, including guidelines regarding valve surgery and device implantation.1
Ejection fraction has limitations imposed by load dependence and technical challenges, especially related to accurate tracing of endocardial borders and assumptions regarding LV geometry.3 Two-dimensional strain (2DS) is a method of quantifying regional tissue deformation from continuous frame-by-frame tracking of acoustic speckles, which is angle independent, less subject to artifact, and simpler to obtain than Doppler-derived tissue velocity imaging.4, 5 Global LV function may be assessed with speckle tracking either from averaging of segmental 2DS or from automated measurement of EF obtained from endocardial detection in end-diastole and end-systole from speckle tracking. We sought to validate the accuracy of average 2DS and automated EF calculation against other echocardiographic means of EF measurement, using magnetic resonance imaging (MRI) as the reference standard.
Methods
Patient selection
We studied 62 patients (mean age 67 ± 10 years, 50 males) with known or suspected myocardial infarction (MI). All patients underwent conventional 2D and 3-dimensional (3D) echocardiography and MRI (median interval 8 days).
Two-dimensional echocardiography
Apical views were obtained by standard 2D echocardiography, using commercially available equipment (Vivid 7, General Electric Medical Systems, Horten, Norway) and using a 3-MHz phased array probe. Grayscale images were obtained at a frame rate between 50 and 70 per second, and the digital loops were subsequently saved onto an optical disc for offline analysis (EchoPac 6.0, General Electric Medical Systems). End-systolic and end-diastolic LV volumes were obtained using the method of discs approach in the apical 4- and 2-chamber views,6 and EF was calculated (2D-EF).
Real-time 3D echocardiography
Real-time 3D echocardiography images were also obtained from an apical window with the patient in the same position as 2DE. Full-volume images were also gathered over 4 cardiac cycles using a matrix array transducer (×4 transducer, Philips iE33, Andover, MA).7
Measurements of real-time 3D echocardiography volumes and EF were performed off-line (4D analysis, Tomtec GmbH, Unterschlessheim, Germany). Endocardial contours were marked in 12 slices (ie, 15° per slice) in the end-diastolic volume and end-systolic volume. Contour tracing was performed with semiautomatic border detection—after first identifying the apex and mitral annulus on each slice, a preconfigured ellipse was fitted to the endocardial borders of each frame and adjusted as required.8 An endocardial shell was similarly defined and the enclosed 3D volume was calculated.
Two-dimensional strain
The 2D acquisition was used for measurement of automated EF (auto-EF) from 2D strain border recognition. This offline analysis involved tracing the LV endocardium in end systole in both the apical 2- and 4-chamber views. The software then tracked endocardial speckles to outline the end-diastolic and end-systolic LV contours and thereby automatically measure volumes, with auto-EF calculated using the method of discs approach; endocardial border traces were repeated or modified in segments with unsatisfactory tracking.
Frame-by-frame tracking of speckles throughout the LV wall over the cardiac cycle also allowed waveforms of segmental strain and strain rate to be calculated.5, 9 After exclusion of waveforms with noise or artifact, the remaining segmental values were averaged off-line to derive GLS. Tracking quality (TQ) was also averaged in basal, mid, and apical segments.
Magnetic resonance imaging
Cardiac MRI was obtained using a Sonata 1.5-T scanner (Siemens, Erlangen, Germany). Left ventricular images were acquired in horizontal and vertical long- and short-axis views using free induction, steady-state precession imaging. All images were acquired during breath-hold. Offline calculation of the end-diastolic volume, end-systolic volume, and EF was performed using a 3D model (CIM 4.2, Cardiac Image Modeling, Auckland University, Auckland, New Zealand). With the use of 2 long-axis and ≥6 short-axis views, markers were placed on the right ventricle and LV annulus, and the endocardial border was detected automatically.10 The same method was used to detect the epicardial border in long- and short-axis views.
Statistical analysis
Correlations were sought between echocardiographic and MRI measurements, and agreement was expressed using the Bland-Altman method. Linear regression was used to convert GLS into a measure of EF. Data analyses were performed with SPSS 15.0 (SPSS, Inc, Chicago, IL). A P value <.05 was considered statistically significant.
Results
Patient characteristics
Of the 62 patients studied (age 67 ± 10, 50 men), all had previous infarction. The regional wall motion score was 1.7 ± 0.5, and wall motion abnormalities were present in the right coronary artery territory in 87%, the left anterior descending in 71%, and the left circumflex in 48%. Image quality was good in 33, fair in 19, and poor in 10 patients.
Feasibility
Across the entire population, global TQ equaled 1.1 ± 0.2, implying that all segments approved for speckle tracking by the 2D strain software were tracked reliably. However, abnormal curves, believed to be artefactual, were excluded through visual assessment (7% of 744 segments). Image quality did not affect the overall accuracy of each modality (Table I), which may reflect the greater time spent on poorer images to achieve the best possible data.
Table I. Influence of image quality on the correlation between MRI-EF and the echocardiographic techniques of EF calculation
| Good (n = 33) | Fair/poor (n = 29) | P (good vs fair/poor) | |
|---|---|---|---|
| GLS | r = 0.67 | r = 0.66 | z = 0.09 |
| P < .0001 | P < .0001 | P = .93 | |
| 3D-EF | r = 0.75 | r = 0.76 | z = 0.12 |
| P < .0001 | P < .0001 | P = .9 | |
| 2D-EF | r = 0.68 | r = 0.53 | z = 1.26 |
| P < .0001 | P < .001 | P = .22 | |
| Auto-EF | r = 0.74 | r = 0.58 | z = 1.52 |
| P < .0001 | P < .0001 | P = .14 |
The measurement time for cardiac magnetic resonance images and 3D echo was 630 ± 60 seconds. Measurement time for 2D-EF was 90 ± 20 seconds, whereas GLS was performed in 132 ± 30 seconds.
Correlation of 2DS and MRI-EF
Global longitudinal strain showed a moderate linear correlation with MRI-EF (r = −0.69, P < .0001) (Figure 1, A), this result being comparable (P = .06) to the correlation of 3D-EF with MRI-EF (r = 0.80, P < .0001) (Figure 1, B). Global longitudinal strain was still a better modality for assessing LV function than 2D-EF (r = 0.58 P < .0001) (Figure 1, C) and the auto-EF (r = 0.62, P < .0001) (Figure 1, D), although the differences in variation were not significant (z = 1.42, P = .16 and z = 0.94, P = .35, respectively). When the analysis was limited to patients with good-quality images (TQ 1), there was better correlation of MRI-EF with GLS (r = −0.73, P < .0001) and auto-EF (r = 0.67, P < .0001).
Figure 1.
Correlation of MRI-ejection fraction (%) with (A) global longitudinal strain (%), (B) 2D-ejection fraction (%), (C) 3D-ejection fraction (%), and (D) automated ejection fraction (%, derived from speckle tracking of endocardial borders).
Calculation of ejection fraction
To convert GLS into an equivalent MRI-EF, we used linear regression to develop the formula: EF = −4.35 * (strain + 3.9). Applying this formula to our patient population, we found GLS greater than −15.3% is associated with a normal EF by MRI. This regression was validated in a separate group of 10 patients with a range of EFs from 21% to 56% as measured by MRI. Correlation analysis showed similar agreement in the validation group (r = −0.74, P = .015) as in the test group.
Of the 32 patients with a normal MRI-EF (≥50%), 75% had normal systolic function by GLS, whereas 85% of patients were recognized as having a normal 3D-EF. Fewer patients were recognized as normal by 2D-EF (70%, P = .14) and auto-EF (61%, P = .04).
Influence of number of abnormal segments on EF estimation
The impact of LV geometry on the measurement of EF was assessed by comparing groups with ≤6 and >6 abnormal segments (Table II). The correlation of MRI-EF with GLS was significantly better in those with >6 abnormal segments (r = −0.77, P < .0001) than in those with ≤6 (r = −0.40, P < .025 [z = 4.58, P < .001]). In contrast, there was no difference in the correlation of 3D-EF with MRI (r = 0.76, P < .0001; r = 0.72, P < .0001). The correlation between both 2D-EF and auto-EF with MRI-EF also improved significantly with an increase in the number of abnormal segments; however, there were no independent predictors to explain this occurrence.
Table II. Influence of the extent of wall motion abnormalities on the correlation between MRI-EF and the echocardiographic techniques of EF calculation
| GLS | 3D-EF | 2D-EF | Auto-EF | |
|---|---|---|---|---|
| ≤6 (n = 34) | r = −0.40 | r = 0.72 | r = 0.30 | r = 0.34 |
| P < .025 | P < .0001 | P < .095 | P < .05 | |
| ≥6 (n = 28) | r = −0.77 | r = 0.76 | r = 0.63 | r = 0.64 |
| P < .0001 | P < .0001 | P < .0001 | P < .0001 |
Influence of image quality on EF estimation
The impact of image quality on the correlation of MRI-EF with various echocardiographic markers is summarized in Table I. No significant differences in correlations were detected.
Discussion
The traditional measurement of EF using 2D echocardiography is based upon tracing the LV borders and calculation of LV volumes using geometric assumptions. The use of the automated EF technique is less dependent on border tracing (although still dependent on image quality), and 3D techniques are independent of geometric assumptions. Global longitudinal strain derived from speckle tracking, to date used mainly for evaluation of regional function, is independent of geometric assumptions as well as less dependent on endocardial visualization. The results of this study show that this technique is an effective method for quantifying LV function.
Automated measurement of LV function
A number of echocardiographic techniques permit automated evaluation of LV volumes and EF, all of which are influenced by loading conditions. Acoustic quantification permits online measurement of volumes and EF, based on the use of myocardial backscatter to recognize the endocardial border.11 However, this technique is limited by image quality as well as image settings.12 Edge detection is also possible using various edge-detection algorithms based on the recognition of LV shape in both 2- and 3-dimensions, and these may also be used to derive volumes and EF.
Doppler measurements of stroke volume are dependent on the accuracy of LV outflow tract measurement,13, 14 errors in the measurement of which are squared in the course of obtaining the area, limiting the reliability of this parameter. The myocardial performance index is reproducible, easily obtainable from the sum of the isovolumic contraction time and isovolumic relaxation time divided by the ejection time, and correlates closely with invasive measures of both systolic and diastolic function.15 Although independent of LV geometry, its connection with EF is not intuitive. Tissue Doppler techniques may be used to measure annular displacement, an analog of EF,16, 17 but although they are less dependent on image quality than 2D imaging, they are acutely influenced by imaging angle. All of these parameters (and indeed probably also global 2DS) share some component of load dependence.
Two-dimensional strain has several advantages over other echo-based techniques for quantifying LV function. As strain and strain rate waveforms are created from automated detection of the endocardium, such a technique is compatible with novice users. Global longitudinal strain also has a potential time advantage over calculating EF by 3D-echocardiography. Furthermore, global LV markers such as GLS can be calculated even in the presence of poor TQ across 1 or 2 segments. However, adequate TQ is a potential limitation in 2D strain analysis, as accurate calculation of global parameters and automated function may be compromised in patients with poor echocardiograms or low frame-rate images. In the case of poor tracing or speckle tracking, automated measurements such as EF may be inaccurate.
Reliability of 2DS
To our knowledge, the accuracy of global strain for calculation of EF has not hitherto been validated. Former investigations have, however, shown 2DS to be an accurate and promising technique in the assessment of global and segmental function.
Regional deformation measured by 2D strain has been validated against sonomicrometry, both in vitro and in vivo. Korinek et al9 demonstrated a significant correlation (r = 0.99, P < .0001) and close agreement between 2DS and sonomicrometry strain measures (bias ± 2SD = 0.7 ± 2.2%) using a tissue-mimicking compressed gelatin block. Linear regression analysis showed significant correlation (r = 0.94, P < .0001) and agreement (−1.1 ± 7.5%) between in vivo 2DS and sonomicrometry obtained from arrays of crystals in the apical anteroseptal and mid-posterior myocardium of 16 open-chested pigs at baseline and after acute ischemia.
Two-dimensional speckle tracking for LV quantification has also been validated against MRI. The reliability of 2D speckle tracking was demonstrated in a clinical study by Amundsen et al.18 In 11 human subjects, 7 with previous infarction, there was good correlation between 2D speckle-tracking strain and strain obtained from MRI (r = 0.87, P < .001). Intra- and interobserver correlation for strain by speckle tracking was 5.2% and 8.6%, respectively, showing a sound level of feasibility for this technique.
In addition to reliability, the other advantage of 2DS is robustness. In a study of nonechocardiographic readers, Reisner et al2 showed this advantage in 27 consecutive patients after MI and 12 consecutive control patients with a normal echocardiogram. There was good linear correlation between the wall motion score index (a parameter of global LV function) and GLS (r = 0.68, P < .0001). Taking a cut-off value for GLS of −21%, it was found that 2DS had 92% sensitivity and 89% specificity for the detection of LV dysfunction in patients post-MI.
Finally, EF may be insensitive as a specific indicator of the performance and nature of myocardial tissue. In this respect, inconsistencies between GLS and EF may partly reflect the limitations of EF.
Conclusions
The quantification of LV function is important in clinical cardiology and has implications for a variety of treatment decisions.19 The accuracy and reliability of 3D echocardiography have been shown in previous studies to correspond closely to that of MRI, and this echocardiographic technique was also the best in this study. However, for circumstances where either the time or resources are insufficient to use 3DE, the results of this study show that GLS is a promising and accurate technique for quantifying global parameters of LV function.
References
- ACC/AHA/ASE 2003 guideline update for the clinical application of echocardiography—summary article: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (ACC/AHA/ASE Committee to Update the 1997 Guidelines for the Clinical Application of Echocardiography). J Am Coll Cardiol. 2003;42:954–970
- Global longitudinal strain: a novel index of left ventricular systolic function. J Am Soc Echocardiogr. 2004;17:630–633
- . Should we be evaluating the ventricle or the myocardium? Advances in tissue characterization. J Am Soc Echocardiogr. 2004;17:168–172
- Two-dimensional ultrasonic strain rate measurement of the human heart in vivo. IEEE Trans Ultrason Ferroelectr Freq Control. 2002;49:281–286
- Two-dimensional strain—a novel software for real-time quantitative echocardiographic assessment of myocardial function. J Am Soc Echocardiogr. 2004;17:1021–1029
- Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr. 2005;18:1440–1463
- Real-time three-dimensional echocardiography using a novel matrix array transducer. Echocardiography. 2003;20:623–635
- Reproducibility and accuracy of echocardiographic measurements of left ventricular parameters using real-time three-dimensional echocardiography. J Am Coll Cardiol. 2004;44:878–886
- Two-dimensional strain—A Doppler-independent ultrasound method for quantitation of regional deformation: validation in vitro and in vivo. J Am Soc Echocardiogr. 2005;18:1247–1253
- Left ventricular mass and volume: fast calculation with guide-point modeling on MR images. Radiology. 2000;216:597–602
- Automated, on-line quantification of left ventricular dimensions and function by echocardiography with backscatter imaging and lateral gain compensation. Am J Cardiol. 1992;70:1200–1205
- Normal values of left ventricular systolic and diastolic function derived from signal-averaged acoustic quantification waveforms: a multicenter study. J Am Soc Echocardiogr. 2003;16:1244–1251
- Recommendations for quantification of Doppler echocardiography: a report from the Doppler Quantification Task Force of the Nomenclature and Standards Committee of the American Society of Echocardiography. J Am Soc Echocardiogr. 2002;15:167–184
- Real-time 3-dimensional Doppler echocardiography for the assessment of stroke volume: an in vivo human study compared with standard 2-dimensional echocardiography. J Am Soc Echocardiogr. 2005;18:1030–1036
- . Left ventricle myocardial performance index derived either by conventional method or mitral annulus tissue-Doppler: a comparison study in healthy subjects and subjects with heart failure. J Am Soc Echocardiogr. 2005;18:1270–1276
- . The relation between mitral annulus motion and ejection fraction: a nonlinear function. J Am Soc Echocardiogr. 2000;13:896–901
- Assessment of left ventricular ejection fraction using long-axis systolic function is independent of image quality: a study of tissue Doppler imaging and M-mode echocardiography. Echocardiography. 2006;23:846–852
- Noninvasive myocardial strain measurement by speckle tracking echocardiography: validation against sonomicrometry and tagged magnetic resonance imaging. J Am Coll Cardiol. 2006;47:789–793
- Feasibility and clinical decision-making with 3D echocardiography in routine practice. Heart. 2008;94:440–445
Supported in part by a grant in aid (G 04B 1475) from the National Foundation, Melbourne, Australia.
The Cardiovascular Imaging Research Group has joint research ventures, including grant funding, from General Electric Medical Systems, Milwaukee, WI.
PII: S0002-8703(08)00820-X
doi:10.1016/j.ahj.2008.08.032
© 2009 Mosby, Inc. All rights reserved.
