OUP user menu

Strain rate imaging in CRT candidates

Ole A. Breithardt, Lieven Herbots, Jan D'Hooge, Piet Claus, Bart Bijnens, Christoph Stellbrink, Andreas Franke, George R. Sutherland
DOI: http://dx.doi.org/10.1016/j.ehjsup.2004.05.007 D16-D24 First published online: 1 August 2004


The recent introduction of cardiac resynchronization therapy (CRT) revived the interest how to evaluate asynchrony. Doppler myocardial imaging (DMI) allows quantification of both regional myocardial motion and deformation with adequate temporal resolution and has been suggested as a method of patient selection for CRT. Timing analysis of regional myocardial motion can identify heart failure patients with left ventricular asynchrony and has been shown to be of prognostic value in CRT patients. However, myocardial motion may be passive due to tethering effects. Strain rate imaging measures the timing and extent of myocardial deformation and provides more reliable information about the sequence of myocardial contraction in the presence of delayed electrical activation. The advantages and pitfalls of the different DMI techniques are discussed.

  • Echocardiography
  • Pacing
  • Strain rate imaging
  • Heart failure
  • Cardiac resynchronization therapy


The presence of delayed electrical activation within the left ventricle (LV) results in contractile asynchrony with possible adverse effects on LV systolic function.1 Late activated segments are pre-stretched by the early activated segments and exposed to a higher effective preload.2 Regional myocardial blood flow increases to compensate for the elevated myocardial oxygen demand.3 Early activated regions begin to contract while large parts of the ventricle are still relaxed and the very late activated segments will not contribute optimally to ejection of blood if thickening continues after aortic valve closure. This results in inefficient energy expenditure and increased global myocardial oxygen demand.4

Those problems are seen at their most extreme in patients with dilated ventricles and left bundle-branch block (LBBB), where LV activation commences in the septum and spreads slowly through the myocardium. In the study from Vassallo et al.5 total endocardial LV activation time in LBBB, defined as the time from the earliest to the latest LV endocardial activation, ranged from 40 to 180 ms (mean 96±36 ms), which is significantly longer that in normal activation (mean 43 ms, range 29–53 ms)6 and comparable to right ventricular (RV) apical pacing.7 LV activation time is longer in patients with ischemic cardiomyopathy and a history of myocardial infarction.7 This can be attributed to the presence of scars, ischemic degeneration of the conduction fibres and an additional delay between activation and contraction in ischemic myocardium. The latest activated LV region in LBBB is variable, but in most patients found in the basal parts of the postero-lateral LV free wall and in the anterior wall.5

Cardiac resynchronization therapy (CRT) aims to reverse these deleterious effects by early stimulation of the intrinsically late activated regions and has gained increasing acceptance as an adjunct therapy in heart failure. It has been shown to improve symptoms of heart failure and functional status in selected patients with NYHA III–IV, QRS>150 ms and low ejection fraction (EF <35%).8,9 However, a significant proportion of patients does not respond to CRT. This has to be attributed to the ill-defined selection criteria, which rely on QRS width as the only marker for asynchrony.

The apparent need for improved patient selection and follow-up has revived the interest for non-invasive quantification of ventricular asynchrony, since resynchronization can only be effective if correctable contractile asynchrony is present. It might even lead to worsening of hemodynamic function if ventricular stimulation is initiated in a chamber without underlying conduction delay or if unfavourable pacing parameters are chosen. Thus, reliable identification of patients with correctable contractile asynchrony is mandatory to improve patient selection. Doppler myocardial imaging (DMI) is particularly appealing for identification of asynchrony, as it quantifies the regional velocities of myocardial motion with high temporal resolution. Based on this velocity data, information about regional myocardial deformation properties can be determined. The following article will discuss the application of DMI for the assessment of ventricular synchrony and its future role in resynchronization therapy.

Principle of CRT

Echocardiographic patient selection and follow-up requires a basic understanding of CRT devices and the implantation technique. Resynchronization is usually obtained by biventricular stimulation with an endocardial lead in the RV apex and a transvenous coronary vein lead for LV epicardial pacing. A posterolateral LV lead position yields the best hemodynamic results in LBBB patients,10 but the individual response may vary widely and some patients may require pacing of the LV anterior or posterior walls. Most clinicians still favour to perform invasive hemodynamic testing to verify the correct lead position, although this procedure is time consuming and associated with additional risks. Epimyocardial lead placement by thoracotomy is an alternative if implantation through the coronary sinus is not feasible.

In the first generation CRT devices, both leads are connected to the same output channel, which allows only simultaneous biventricular stimulation. New devices separate the output channels and allow selective pacing at each site and the introduction of an interventricular delay between RV and LV stimulation, which might be of additional benefit in some individuals and which might be used to compensate for suboptimal anatomic lead positions (see below).

CRT has demonstrated effectiveness in patients with stable sinus rhythm8 and is preferably performed in an atrial triggered pacing mode (VDD) in this group. Atrio-ventricular (AV) delay optimization contributes to hemodynamic improvement in CRT patients by optimizing LV filling pressures11 and reducing pre-systolic mitral regurgitation in patients with long PR intervals. Conventional blood pool Doppler imaging is the most widely studied modality for noninvasive AV-delay optimization12,13 and it has been shown that Doppler guided AV delay optimization is associated with a temporary or medium-term improvement in symptoms and exercise capacity.14 Modification of the AV delay in CRT leads to characteristic changes in Doppler parameters,15 but prospective results on non-invasive optimization in CRT is not yet available.

The experience with DMI for AV delay optimization is limited. Gessner et al.16 compared PW-DMI measurements at the basal septum with stroke volume estimations by CW Doppler during RV apical pacing. They concluded that the time interval between the end of the late diastolic filling wave (A) and the onset of regional systolic motion should be 70–80 ms to obtain the optimal AV delay. Although this interval is similar to normal values, it is unclear, whether the results can be transferred to the CRT population with LBBB.

Beyond the optimization of AV synchrony, the programmed AV delay may also have an impact on ventricular coordination, which is the result of combined “electrical wavefronts”: the intrinsic excitation wavefront through the AV node and the artificially stimulated contraction wavefront, originating at the ventricular pacing sites. If a long AV delay is programmed which is close to the patients PR interval, the ventricle will be mainly activated through the intrinsic pathway. In contrast, short AV delay settings may result in complete preexcitation.

All of the described parameters have impact on the long-term success of CRT. Lead position has to be selected during implantation, but AV delay and – in recent devices – the interventricular delay can be adjusted during follow-up to “titrate the correct dosage” of CRT and might partly compensate for a suboptimal lead position.

DMI techniques for identification of asynchrony and monitoring of CRT

Most investigators and clinicians still rely solely on the electrocardiogram to identify the presence of correctable asynchrony. A QRS cut-off of 130 ms is accepted to select suitable patients for CRT17 and it has been suggested to monitor the change in QRS width to confirm the efficacy of (biventricular) CRT.18 Although these concepts are supported by the clinical results for the majority of patients, a significant proportion of patients does not benefit in the long-term.8,19

Various reasons might be responsible for this therapy failure. If CRT is initiated in a ventricle without underlying asynchrony, any LV pacing site apart from the his bundle will result in worsening of systolic function.20–22 Thus, reliable identification of non-responders necessitates the direct quantification of ventricular contractile synchrony. Among the proposed imaging techniques for quantification of regional asynchrony, echocardiography has the advantage to be widely available and to be easily applicable at the patients bedside.23–25

Traditionally, asynchrony is identified by visual comparison of regional wall motion which is subjective and associated with a significant intra- and interobserver variability.26 Most importantly, the temporal resolution of the trained human eye is limited and regional delays of less than 70 ms will be missed by most examiners.27 Quantitative techniques with high temporal resolution such as DMI overcome this limitation. Information on the velocity of regional myocardial motion can be obtained either by the range-gated pulsed-wave Doppler (PW-DMI) or by the colour-Doppler technique (CDMI), where the Doppler information is color-coded and superimposed on the 2D image. The pulsed-wave technique with its higher temporal resolution (∼4 ms) measures the maximal instantaneous velocities; whereas the CDMI information is obtained by autocorrelation techniques which measures rather the regional mean velocities and results in lower frame rates between 75 and 100 fps (∼15 ms temporal resolution). Recent high-end scanners are able to generate frame rates up to 240 fps with optimal settings (narrow sector angle, low PRF), however at the cost of impaired spatial resolution. But despite the intrinsically lower frame rates, data acquisition with CDMI has several advantages. CDMI contains information on all myocardial segments in the imaging plane and the acquisition process is less operator dependent than sampling several sites with PW-DMI. It allows direct comparisons of several segments in the same cardiac cycle, which is most important for evaluation of asynchrony in CRT. All velocity information obtained by CDMI is stored digitally and enables automated quantification of regional myocardial velocities and post-processing for myocardial deformation imaging (strain rate imaging).

DMI in normal conduction, LBBB and CRT

Tissue velocity imaging

Electrical activation spreads rapidly through the normal conduction system and results in a near synchronous onset of myocardial contraction throughout the ventricle. The average time delay between the onset of the QRS and the onset of regional systolic motion by CDMI is about 100–110 ms. Earliest onset of longitudinal systolic motion is usually observed in the posterobasal segment with a short delay in the other walls. Overall wall motion in the normal ventricle is synchronous throughout the cardiac cycle. Least intersegmental variation is observed in the onset of systolic motion (Fig. 1, open arrow), which can usually be identified as the first zero crossing of the velocity trace after the isovolumic signal.

Fig. 1
Fig. 1

Color Doppler velocity profiles from the basal, mid and apical segments of the septum and the lateral wall in a normal heart, displaying regional wall motion. Systolic (S), early (E) and late diastolic (A) myocardial motion is synchronous. Open arrow indicates onset of systolic motion.

For LBBB patients, the timing of systolic motion onset has been tested in the assessment of mechanical asynchrony between the RV and LV free walls.28 In this study, QRS width was closely associated with an increased delay in the onset of systolic motion between the right and the LV as measured by PW-DMI in the basal segments of the ventricular free walls. Echocardiography was of particular value in patients with only modest QRS prolongation between 120 and 150 ms. In this subgroup, DMI identified patients with a higher degree of asynchrony than expected from QRS width alone. It has to be stressed that this analysis was focusing on inter ventricular synchrony (i.e. between the RV and the LV) and that the septum was not analysed, thus, no conclusions can be made on left intra ventricular synchrony. The effects of CRT on the onset of regional systolic motion has also not been tested yet.

An alternative approach to assess wall motion synchrony is to compare the timing of peak systolic velocities. However, it has to be noticed that timing of peak systolic velocity may demonstrate significant intersegmental differences even in the normally activated ventricle. The systolic peak of the velocity curve is sometimes flattened and the lateral wall velocity tracings frequently show a biphasic pattern – both phenomena may complicate the identification of peak systolic velocity. In diastole, peak velocity during early filling (peak E) occurs earliest in the basal segments and latest in the apical segments and is dependent on the flow propagation velocity during early diastole.

The effects of CRT on the timing of regional peak systolic velocities in six basal and six mid LV segments was studied by Yu et al.29 A large intersegmental variation in this interval, measured from QRS onset to the peak systolic velocity (Ts) was found in LBBB, ranging from 148±25 ms at the basal anteroseptal segment to 216±52 ms in the basal lateral segment. Resynchronization with biventricular pacing resulted in a more homogeneous distribution with a reduction in the intersegmental variation, ranging after CRT from 191±32 ms (basal anteroseptal) to 213±44 ms (basal lateral). The resulting standard deviation of (Ts-SD) was used as a measure for intraventricular dyssynchrony and was significantly reduced from 37.7±10.9 ms during intrinsic LBBB to 29.3±8.3 ms during CRT. Later the same group evaluated the predictive value for Ts-SD in a small group of 30 patients and proposed a cut-off value for baseline Ts-SD of 32 ms to identify CRT responders, defined as patients who show reduced LV volumes (reverse remodelling) during long-term CRT.

New parametric imaging techniques have been developed to facilitate the time-consuming measurement of Ts. Tissue Synchronization ImagingTM (GE Vingmed, Horten, Norway) enables online identification of delayed peak systolic motion by color-coding and allows quick measurements of Ts and regional peak systolic velocities. However, reproducibility and reliabilty of this new parametric imaging technique still has to be evaluated. The automatic measurement is dependent on the correct identification of LV systole. In the current software, this time interval is preset to a standard interval adapted to the cycle length, which may not identify precisely enough the onset of LV systole (ejection) in all patients. Thus, it remains necessary to verify in each patient the correct time interval settings, by the comparison to global cardiac events (timing of valvular opening and closure).

Tissue tracking

Based on the DMI velocity data, information on long-axis myocardial displacement can be derived from the integration of myocardial velocities and displayed as a 2D color map, the so-called `tissue tracking' image. The information presented in such a color map has been shown to correlate well with the presence of dobutamine stress induced regional wall motion abnormalities and might help to improve the interobserver variability in the interpretation of stress echocardiography results.30 The technique may also allow to estimate regional shortening in a segment, i.e. regional strain.

Tissue tracking has been applied in CRT patients to identify regional systolic wall motion abnormalities during intrinsic conduction and to display the effects of CRT.31,32 In this single-center experience, the visual analysis of end-systolic tissue tracking color maps identified segments with post-systolic apical displacement. The presence of post-systolic motion, in these studies defined as delayed longitudinal contraction (DLC), was verified by an additional strain rate analysis (see also next section) and correlated to the long-term success of CRT.

However, the application of tissue tracking for the quantification of myocardial asynchrony has several methodological limitations. Calculation of regional apical displacement commences at the automatically identified R-wave and ends at a user-selected time point, usually the end of the T-wave.This interval does not always represent the true regional contraction period, particularly not in the presence of an LBBB. The tissue tracking algorithm does not differentiate between the sustained velocity peak of systolic motion and the short-lived isovolumic velocity spike. Regions with either no displacement or with negative displacement will both show the same tissue tracking information (no color). The tissue-tracking color-bars may differ markedly and display misleading information, depending on the chosen interval duration (Fig. 2). A possible solution might be, to define the systolic interval objectively by identification of the global systolic events (aortic valve opening and closure).

Fig. 2
Fig. 2

Tissue tracking images from the same cardiac cycle in the apical four-chamber view. Vertical dotted lines indicate the selected interval for the integration of velocity data. The left image shows the resulting color-map with a short systolic period, where the end of systole was set before the end of the electrocardiographic T-wave. The apical septum appears gray, indicating lack of systolic apical displacement. In the right image, endsystole was defined later, after the end of the T-wave. With this setting, the whole septum shows some degree of apical displacement, while the apical segment of the lateral wall appears gray.

Myocardial deformation imaging

To overcome the limitations of velocity imaging, CDMI has been further developed to quantify regional myocardial deformation.33,34 Identification of regional deformation properties is of particular importance to identify the ideal CRT candidate. This information can be obtained from CDMI velocity data sets by calculating the instantaneous velocity gradient between two samples with a predefined fixed distance. The result is divided by the sample distance (typically 10 mm) and yields the temporal change of deformation, myocardial strain rate (SR). Integration of SR yields the amount of regional deformation over time, myocardial strain (Math, %). Myocardial deformation imaging (∼SR imaging, SRI) provides valuable additional information on regional function and has been used for identification of regional ischemia35 and viability.36

In comparison to the analysis of myocardial velocities, the SR approach better reflects the active deformation properties (lengthening and shortening in the longitudinal axis). Only deformation imaging is able to identify reliably the presence of delayed contractions and thus the presence of a recruitable contractile reserve that can be utilized by CRT by transferring it from the pre or post-ejection period to the systolic ejection phase (Figs. 3 and 4). A patient in whom active contraction in an early activated region (e.g. the septum) is not followed by late- or post-systolic contraction in a remote region, will probably not benefit from pacing, as there is no recruitable contractile reserve. Patients with coronary artery disease might represent an exception, as post-systolic deformation is also determined by the presence of regional ischemia. The extent of post-systolic deformation in ischemic segments is not only dependent on the time to regional activation, but also on the presence and level of ischemia. In ischemic segments a delay from activation to contraction has been observed and this might represent an independent level of asynchrony.37 In this setting, early stimulation by CRT might shorten the time to regional activation, but there might still remain significant post-systolic deformation due to the ischemic activation-contraction delay.

Fig. 3
Fig. 3

Comparison between the onset of regional radial motion and strain in the postero-lateral Graphic, posterior Graphic and postero-septal Graphic segments in a patient with LBBB (parasternal short axis view). A marked inter-regional delay to the onset of regional thickening (positive strain, right column) can be observed, despite small heterogeneity in the onset to regional motion (velocities, left column). All segments show delayed thickening after aortic valve closure (∼post-systolic thickening).

Fig. 4
Fig. 4

Post-processed velocity and strain curves before (LBBB) and after CRT from the septal and lateral wall mid segments in a 73-year-old patient with idiopathic dilated cardiomyopathy. The derived strain curves from the mid segments clearly show the regional asynchrony in maximal deformation. Maximal septal contraction occurs before aortic valve opening (left, arrow) and during lateral wall relaxation. Peak lateral wall contraction is observed very late in systole and persists into the post-systolic period (left, arrowhead). With resynchronization, systolic contraction occurs simultaneously in both walls and both segments contribute equally to ejection (right). Note also the shortening of the isovolumic contraction time (IVCT) with CRT. Isovolumic relaxation time (IVRT) is less affected by CRT.

Strain rate imaging has been applied in CRT patients to identify delayed segmental activation and contraction.31 DMI velocity traces from the LV base were analyzed to identify post-systolic anterior motion (positive velocity signal during IVRT). In such segments with delayed motion, an additional SR analysis was performed to quantify regional deformation properties. The authors assumed that a negative SR indicates active contraction in this segment and demonstrated that the extent of delayed longitudinal contraction at the LV base predicted the long-term efficacy of CRT.

In a recent study with acute reprogramming of the CRT device to no pacing and to active biventricular pacing, we compared regional strain patterns between the septum and the lateral wall and could demonstrate the high prevalence of asynchronous deformation in LBBB.38 In most patients a characteristic pattern with early septal and delayed lateral wall shortening was present during LBBB and resynchronized during active CRT. The quantitative analysis of the regional deformation properties showed that CRT reduced the abnormal stretch in the late activated segments and led to a redistribution of regional systolic strain, suggesting unloading of the lateral wall during LV preactivation. These observations are in accordance to previous experimental work39,40 and might explain some of the beneficial long-term effects of CRT on the regional level. The reduced wall stress during CRT in the previously delayed activated and overloaded posterolateral segments may normalize the abnormal myocardial calcium handling41 and protein dysregulation42 and reduce arrhythmogenecity.43

Another important finding from this acute experiment was the dissociation between the timing of regional motion and deformation.38 In the normal heart, the onset of regional shortening in the longitudinal direction coincides with the peak systolic SR and the peak systolic velocity (Fig. 5, red line). Motion and contraction are tightly coupled and the timing of regional contraction can be estimated from the velocity measurements. In contrast, a dissociation between motion and contraction is observed in LBBB, where myocardial motion is to a large degree affected by neighbouring segments and overall heart motion (Fig. 6). In this setting, velocity measurements provide no reliable information about the timing of regional contraction as the onset of regional motion may significantly preceed the onset of deformation and vice versa.

Fig. 5
Fig. 5

Comparison between regional myocardial velocities by CDMI (upper row), post-processed strain rate (middle row) and strain (lower row) curves in the basal septum and the basal lateral wall in a normal heart. Peak systolic motion is indicated by the red line and occurs simultaneous to the peak strain rate and the onset of regional strain. Strain rate and strain values are negative in systole, indicating regional shortening.

Fig. 6
Fig. 6

Effect of CRT on mid septal (yellow) and mid lateral (green) regional velocities, strain rate and strain in a 65-year-old male with idiopathic dilated cardiomyopathy and LBBB before (OFF, middle column) and after CRT (right column). Regional systolic peak values are indicated by the arrows. Solid arrow represents the septum, dashed arrow the lateral wall. Aortic valve closure is indicated by the dashed vertical line. Before CRT, the peak systolic velocity of the lateral wall occurs before the septal peak, while regional shortening occurs earlier in the septum, as indicated by the peak strain rate and strain. Post-systolic shortening is observed in the lateral wall. CRT reverses the timing of regional shortening and eliminates post-systolic shortening in the lateral wall. Timing of peak systolic velocities is less affected.


The assessment of myocardial asynchrony has regained importance with the recent introduction of resynchronization therapy in heart failure. Improved patient selection for CRT requires thorough identification of myocardial contractile asynchrony. DMI allows precise evaluation of longitudinal wall motion with high temporal resolution, but the application of DMI in heart failure with bundle-branch block is much more challenging than in hearts with preserved systolic function. Technical limitations come to the fore as the hearts are larger and myocardial function (and motion) is reduced. Direct comparison of motion synchrony between two opposing walls requires larger sector angles due to the spherical ventricular enlargement. This limits the available frame-rates and thus temporal resolution. Reduced systolic function results in lower regional myocardial velocities and lower signal to noise ratio, necessitating optimized imaging settings and demanding higher operator skills to obtain interpretable and reproducible data sets.

However, the temporal analysis of myocardial motion by velocity imaging allows no definite conclusions about the timing of regional contraction. Myocardial deformation imaging (SR imaging) should be able to overcome these limitations of velocity imaging in the future and is expected to provide a better estimate of both the extent and the timing of regional contraction. To that end, it is important to study regional interaction with remote segments to verify the presence of active contraction. This information may help to identify responder patients and to guide follow-up.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
  32. [32]
  33. [33]
  34. [34]
  35. [35]
  36. [36]
  37. [37]
  38. [38]
  39. [39]
  40. [40]
  41. [41]
  42. [42]
  43. [43]
View Abstract