OUP user menu

Assessment of left ventricular function by different speckle-tracking software

Ana Manovel, David Dawson, Benjamin Smith, Petros Nihoyannopoulos
DOI: http://dx.doi.org/10.1093/ejechocard/jep226 417-421 First published online: 27 February 2010

Abstract

Aims Two-dimensional (2D) speckle echocardiography enables objective assessment of left ventricular function through the analysis of myocardial strain, which can be measured by different speckle-tracking software. The aim of this study was to compare two different commercially available cardiac ultrasound systems and their manufacturer-specific speckle-tacking software for the quantification of global myocardial strain in a healthy population.

Methods and results Twenty-eight healthy subjects (age: 38 ± 12, 64% males) underwent two 2D echocardiograms within the same day using different cardiac ultrasound systems: Vivid 7 (GE Ultrasound, Horten, Norway) and Artida 4D (Toshiba Medical Systems). Standard apical and short-axis views of the left ventricle were obtained in each subject with a frame-rate range of 60 ± 20 frames/s. Global longitudinal, radial, and circumferential strain values were analysed using their respective speckle-tracking software for Vivid (2D-strain EchoPac PC v.7.0.1, GE Healthcare, Horten, Norway) and Toshiba systems (2D Wall Motion Tracking, Toshiba Medical Systems). Global strain values were estimated from the average of regional left ventricular strain values. Agreement between the two systems and software was assessed by Bland–Altman method. Mean left ventricular ejection fraction was 59 ± 7%. Global longitudinal, radial, and circumferential strain values were, respectively, −21.95 ± 1.8, 46.97 ± 5.5, and −23.18 ± 3.3% when using 2D-strain EchoPac and −22.28 ± 2.1, 40.74 ± 4.3, and −27.17 ± 4.7% when 2D Wall Motion Tracking was used (P = NS). Limits of agreement between both speckle-tracking software were narrower for global longitudinal strain (−2.25 to 3.65) than for radial and circumferential strain (−2.23 to 12.44 and −1.36 to 10.54, respectively).

Conclusion Two commercially available speckle-tracking software appear to be comparable when quantifying left ventricular function in a healthy population. Global longitudinal strain is a more robust parameter than radial and circumferential strain for the assessment of myocardial function when different cardiac ultrasound systems are used for analysis.

  • Left ventricular function
  • Speckle tracking
  • Myocardial strain

Introduction

The advent of two-dimensional speckle echocardiography (2DSE) has allowed a comprehensive and quantitative assessment of left ventricular function.13 On the basis of a frame-to-frame tracking of myocardial acoustic markers and the subsequent spatial and temporal image processing, current 2DSE software enable the analysis of myocardial strain in its longitudinal, circumferential, and radial dimensions.4 Over the past 5 years, numerous studies have demonstrated that impaired myocardial strain is an indicator of global and regional dysfunctions in a variety of cardiac conditions, although different speckle-tracking software have been commonly used.5,6

The accuracy and rapid analysis of 2DSE convert this technique into a suitable method for objective assessment of myocardial function in clinical practice. Nonetheless, consistency of reported data and reference values are needed before its clinical application can be widely accepted. Vendor diversity of currently available speckle-tracking software leads, therefore, to investigate whether quantitative values obtained from every system are similar. Scarce studies have been performed in the past to assess the comparability of systems for tissue Doppler imaging,7 and no data have been reported using 2DSE.

We, therefore, sought to examine the consistency of 2D speckle-derived myocardial strain values in healthy subjects when analysed by two different cardiac ultrasound systems and their respective software.

Methods

Study population

Thirty-three subjects (age: 38 ± 12, 64% males) with the absence of structural cardiac disease were included in the study. All participants underwent two transthoracic echocardiograms within the same day using different cardiac ultrasound systems. Global longitudinal, radial, and circumferential strain values were analysed in all subjects using the specific 2D speckle-tracking software of each system.

Echocardiography

Transthoracic echocardiograms were performed using two commercially available ultrasound transducer and equipment (M3S probe, Vivid 7, GE Healthcare, Horten, Norway; 3 MHz/PST-30SBT probe, 4D Artida, Toshiba Medical Systems). Standard three apical and short-axis views of the left ventricle were obtained using a frame-rate of 40–80 frames/s. Three cardiac cycles at each plane were stored in cine-loop format in order to subsequently select the images of better quality for off-line speckle-tracking analysis. The same transducer frequency (1.7–3.6 MHz) was used in both systems, and special care was taken to use similar sector width, focus position, and frame-rate range during image acquisition.

Two-dimensional speckle-tracking analysis

Speckle-tracking analysis was performed by the respective customized 2DSE software for Vivid (2D-strain EchoPac PC v.7.0.1, GE Healthcare, Horten, Norway) and Toshiba system (2D Wall Motion Tracking, Toshiba Medical Systems). The endocardial border was manually traced, and a region of interest was drawn to include the entire myocardium in all cases. The software algorithms automatically segmented the left ventricular planes into equidistant segments and then performed speckle tracking on a frame-to-frame basis. When using 2D-strain Echopac, images were accepted for analysis when global tracking quality index equalled 1, which implied that all segments approved for speckle analysis were tracked reliably. When 2D Wall Motion Tracking was used, exclusion of the studies was done upon visual assessment, when abnormal curves were believed to be artefactual. Longitudinal strain values were obtained from the three apical views, and radial and circumferential strain measurements from the basal, mid, and apical short-axis planes. Global strain values were calculated from the average of regional values and stated in per cent (%). Intraobserver and interobserver reproducibility of longitudinal, radial, and circumferential strain values was analysed with repeated measurements by the same observer at two different time points and by a second independent observer in five randomly selected subjects. Both observers were blinded to the results of the other software package and previous strain results when assessing reproducibility.

Statistical analysis

All values were expressed as mean ± SD. Agreement between the two speckle-tracking methods was assessed by Bland–Altman analysis.8 The mean difference and limits of agreement between the measurements derived from each system were calculated. Intraobserver and interobserver variability values were estimated as the absolute difference between the corresponding two measurements in per cent of their mean in each patient and then average over the five patients. Statistical analysis was performed using SPSS software (version 15.0, SPSS Inc, Chicago, IL, USA).

Results

Baseline characteristics

All study subjects showed normal systolic function with LVEF, fractional shortening, and myocardial systolic velocities by tissue Doppler within normal limits. Diastolic function, as assessed by mitral valve inflow pattern and E/A ratio at the lateral mitral annulus, was also normal in all subjects (Table 1). Average heart rate showed no differences when echocardiogram was performed with both ultrasound systems (64 ± 11 vs. 68 ± 12 bpm, P = NS).

View this table:
Table 1

Clinical and echocardiographic characteristics of study patients (n = 28)

Age (years)38 ± 12
Male (%)18 (64)
LVEF (%)59 ± 7
FS (%)33 ± 4
Sa (cm/s)10 ± 2
E/A1.5 ± 0.4
Ea/Aa2.0 ± 0.6
  • LVEF, left ventricular ejection fraction; FS, fractional shortening; Sa, systolic tissue velocity at lateral mitral annulus; E/A, ratio of early and late diastolic waves of mitral inflow; Ea/Aa, ratio of early and late diastolic tissue velocities at lateral mitral annulus.

  • LVEF, left ventricular ejection fraction; FS, fractional shortening; Sa, systolic tissue velocity at lateral mitral annulus; E/A, ratio of early and late diastolic waves of mitral inflow; Ea/Aa, ratio of early and late diastolic tissue velocities at lateral mitral annulus.

Speckle-derived strain analysis

Speckle analysis was feasible in 28 subjects from initially recruited 33, since 5 were excluded for speckle analysis due to inadequate tracking by either of the software. Time required for speckle analysis and interpretation of all strain curves was 7 ± 2 and 9 ± 3 min for Vivid and Toshiba systems, respectively.

Similar peak global longitudinal strain was found when 2D-strain EchoPac and 2D Wall Motion Tracking software were used (−21.95 ± 1.8 vs. −22.28 ± 2.1%, respectively; P= 0.56). Representative examples of longitudinal strain measured by the two systems are shown in Figure 1. For measurements derived from short-axis planes, radial strain appeared to be lower (46.97 ± 5.5 vs. 40.74 ± 4.3%, P = 0.19) and circumferential strain higher (−23.18 ± 3.3 vs. −27.17 ± 4.7%, P = 0.06) when using Toshiba system than those values obtained from 2D-strain EchoPac. Nevertheless, neither of them reached statistical significance.

Figure 1

Example of two-dimensional speckle-derived longitudinal strain curves obtained from apical four-chamber view using 2D-strain EchoPac (Vivid) (A) and 2D Wall Motion Tracking (Toshiba) (B) software. Images on the left side represent the region of interest included for speckle analysis, which is automatically divided into six segments and codified in different colours by either system. Graphs on the right side show global and regional longitudinal strain curves. Global values are depicted as white lines (dotted line in A and solid line in B) and regional strain curves in different colours corresponding to each myocardial segment.

The comparability of global strain measurements between both speckle-tracking software showed good agreement for global longitudinal strain, with an average difference of 0.33 ± 1.48 (mean ± 2 SD) and limits of agreement of −2.25 to 3.65. The limits were found to be wider for radial and circumferential strain (−2.23 to 12.44 and −1.36 to 10.54, respectively) than for longitudinal strain measurements. Bland–Altman plots are depicted in Figure 2.

Figure 2

Bland–Altman plots represent the agreement between 2D-strain EchoPac (Vivid) and 2D Wall Motion Tracking (Toshiba) measurements for global longitudinal, radial, and circumferential strain. Dotted horizontal lines denote the bias (mean inter-technique difference) and solid horizontal lines depict the 95% limits of agreement (2 SD around the mean difference). GLS, global longitudinal strain; GRS, global radial strain; GCS, global circumferential strain. Sub-index a: Vivid system; sub-index b: Toshiba system.

Intraobserver and interobserver variability in the strain measurements was <6 ± 2 and 6 ± 4% for Vivid system, and <8 ± 9 and 7 ± 11% for Toshiba system, respectively.

Discussion

The main results of this study are that two different commercially available software show good comparability for global longitudinal strain as a parameter of overall left ventricular function, and that lower agreement is found for myocardial deformation parameters derived from short-axis views than for longitudinal strain. These findings are important in view of future clinical applications of 2DSE, particularly for patient diagnosis and follow-up in centres where diversity of cardiac ultrasound systems exist. To our knowledge, this is the first study which compares the 2DSE-derived myocardial strain measurements between different available software.

Advantages and pitfalls of speckle-derived strain measurements

Owing to the angle independency tracking, 2DSE enables to assess myocardial deformation in 3D and at different cross-sectional views throughout the left ventricle, which provides more comprehensive assessment of myocardial function than Doppler-based modalities. Previous studies have shown that the normal myocardium exhibits ∼20% of global deformation in its longitudinal dimension as assessed by 2DSE.4,9 In our study, the value of global longitudinal strain obtained in healthy subjects is the same as reported previously. In addition, the present research shows that global longitudinal strain measurement by 2DSE seems to be consistent when two different 2D-strain software are used for analysis.

Strain measurements derived from short-axis views have demonstrated, however, less agreement than longitudinal strain values in this subpopulation. Nonetheless, it is important to mention that when global radial and circumferential measurements were compared with those obtained in previous studies using the same software, similar values were found.10,11 The observed variation between software when radial and circumferential strain are analysed might be somehow expected owing to the limitations inherent to 2D speckle-tracking. Two-dimensional speckle echocardiography is highly influenced by 2D image quality, and tracking in the lateral direction is particularly dependent on lateral resolution. The fact that speckles are tracked almost within the same scan line when longitudinal strain is assessed might explain the high reproducibility of this parameter. Radial and circumferential strain measurements are, on the other hand, more influenced by the line density and, therefore, highly subjected to the appropriateness of parameter setting for image acquisition. It is also important to note that the considerable out-of-plane motion of acoustic markers in short-axis views could also represent a limitation for the consistency of findings when transmural and circumferential motion are analysed. In addition, the presence of a transmural deformation gradient and differently oriented myofibre layers within the left ventricular wall should be kept in mind,12,13 since the initial position from where myocardial acoustic markers start to be tracked might differ between software. When validated against cardiac magnetic resonance, transmural strain and circumferential strain by 2DSE have also been reported to have only a fair correlation.14,15

Clinical relevance of the findings and comparability between systems

Since the advent of new cardiac imaging technology based on tissue characterization, the comparison of values from different software manufacturers has been scarcely reported. A previous study using tissue Doppler imaging showed that different systems have sufficient accuracy and agreement for the estimation of myocardial velocities, although variations in measurements of strain and strain rate were present.7 Two-dimensional speckle echocardiography employs different calculation algorithms from Doppler-based modalities, and whether variations on strain measurements by 2DSE exist was unknown. This is the first study comparing 2D speckle analysis by two different systems. The fact that we have obtained a good agreement and reproducibility for longitudinal strain measurements between EchoPac and 2D Wall Motion Tracking software suggests that this parameter could be suitable for the assessment of global left ventricular function independently whether Vivid 7 (GE Healthcare) or Artida 4D (Toshiba Medical Systems) cardiac ultrasound system is used for analysis. This statement is important, as longitudinal myocardial fibres are known to be first affected in many cardiac conditions.16,17 Moreover, long-axis function constitutes a substantial component of left ventricular performance, despite the preservation of motion derived from circumferentially oriented myofibres and left ventricular ejection fraction. The findings of this study might be the first step in the searching of conventional strain values for different cardiac conditions, allowing the use of different software available in each institution. This is of notable interest in view of the future clinical application of speckle-tracking echocardiography. With regard to radial and circumferential strain, the clinical significance of the maximal inter-technique difference observed is to be clarified before accepting the use of either speckle-tracking software for the same clinical purpose.18

Study limitations

The results of this study have been obtained from a small group of healthy volunteers. Therefore, extensive validation studies in different clinical settings are necessary in order to be able to extrapolate conclusions to the variety of cardiac conditions. This study could be the guide of future research in this particular setting.

Owing to the dependency of 2D speckle analysis on image quality, special care was taken to keep to a minimum the difference in parameters setting between systems when acquiring the images. However, some inter-technique variations were still present. Of note, the consideration of a frame-rate range rather than an exact frame-rate value in the studies and the impossibility to keep the same region of interest thickness between systems are some of the aspects which could have affected our results. Finally, although the algorithms used by the different software packages are supposed to be similar in terms of the speckle-tracking analysis, which is based on a block-matching approach of the speckle patterns within the myocardium, and the strain calculation as the relative change in length between individual points (ε = L(t) − L0)/L0), some software-inherent differences in data processing might exist.

Conclusion

Two different commercially available cardiac ultrasound systems appear to be comparable when quantifying left ventricular function by 2DSE-derived global longitudinal strain. Until more research is available, strain values obtained from short-axis views should be considered more cautiously in the assessment of myocardial function when different speckle-tracking software are used for analysis. Further, similar studies on different cardiac conditions are needed in order to be able to provide reference values for the clinical application of 2DSE.

Funding

The first author was recipient of a grant from the Spanish Society of Cardiology.

Conflict of interest: none declared.

References

View Abstract