Wall motion tracking is an application of pattern-matching ultrasound commonly known as speckle-tracking echocardiography (STE). STE involves creating a template image using a local myocardial region in the starting frame of the image data. In the ensuing frame, an algorithm searches for the local speckle pattern that most closely matches the template. A movement vector is then created using the location of the template and the matching pattern in the subsequent frame. Multiple templates are used to observe movement of the entire myocardium. The process is then repeated by creating new templates and observing their movements in subsequent frames until the entire cardiac cycle has been assessed.
Cardiac performance is a 3D phenomenon; therefore, 2D tracking is intrinsically limited. 2D tracking follows motion occurring within the imaging plane, while the out-of-plane motion component results in noise and interferes with tracking. The inability of 2D STE to measure one of the three components of the local displacement vector is an important limitation that affects the accuracy of the derived indices of local dynamics.
Following the introduction of 3D STE by Abe et al.1 in 2008, Maffessanti et al. compared the segmental 2D and 3D STE values of six indices (radial and longitudinal displacement, rotation and radial strain [RS], longitudinal strain [LS] and circumferential strain [CS]);2 these values did not appear to correlate well, with wide variations of inter-technique agreement. The authors took a variety of 3D STE measurements of left ventricular (LV) regional wall motion and demonstrated their advantages over 2D STE measurements. Segments were classified as normal or abnormal using cardiac magnetic resonance imaging. In normal segments, 3D STE showed higher displacements, due to the out-of-plane motion component of 2D STE, and smaller standard deviations, indicating that tighter normal ranges were found with 3D. Furthermore, gradual decreases in displacement as well as reversals in rotation from base to apex were demonstrated with 3D STE. In abnormal segments, all 3D STE indices were reduced. Thus the relatively smaller normal variability of 3D STE compared with 2D STE measurements provided additional support for the superiority of 3D STE in handling the complexity of regional wall motion abnormality. While 3D STE generates over 3,000 vectors per volume and its temporal resolution is the same as the frame rate of realtime 3D data sets (20–30 volumes/second), it actually reduces the examination time to one-third that of 2D STE.3,4
The potential benefits of the 3D STE technique are evident from a recent comparison between 3D STE and sonomicrometry in anaesthetised sheep. Segmental LS, RS and CS were measured at baseline, during pharmacological stress (induced by a dobutamine and propranolol infusion) and acute myocardial ischaemia (induced by coronary artery occlusion).5 Good correlations were observed between measurements obtained by 3D STE and those obtained by sonomicrometry (LS: Pearson correlation coefficient [r]=0.89, p-value [p]<0.001; RS: r=0.84, p<0.001; CS: r=0.90, p<0.001). In each segmental study, significant correlations of the three strain components were observed (LS: r=0.65–0.68, p<0.001; RS: r=0.59–0.70, p<0.001; CS: r=0.71–0.78, p<0.001).
3D STE was also tested in vitro using a twist phantom, with the heart base rotating at 0°, 15°, 20° and 25° along a fixed apex to avoid translational motion.6 Segmental and global rotation at basal, middle and apical segments correlated well with the actual rotation (base: r=0.93, middle: r=0.92, apex: r=0.95, all p<0.001). To define the percentage change in a segmental area during systole, the area change ratio (ACR) obtained by area tracking was calculated using 3D STE versus sonomicrometry at baseline, during pharmacological stress and acute myocardial ischaemia induced by occlusion of the mid-left ascending artery.7 A strong correlation was found between ACR measurements obtained by 3D STE and those obtained by sonomicrometry (r=0.87, p<0.001).
Despite experimental data, we continue to have no true non-invasive ‘gold standard’ technique that can be used in humans to validate regional ventricular function in three dimensions. Nevertheless, 3D STE has been evaluated in a variety of clinical scenarios to determine LV volume8–11 or acute and chronic ischaemic heart disease.12–16 Compared with normal individuals, patients with hypertrophic cardiomyopathy (HC) have been shown to have increased peak LV twist (16.5 ± 4.7° versus 12.0 ± 3.9°, p<0.001), predominantly because of increased apical rotation in those with LV outflow tract obstruction (patients with obstruction 12.7 ± 4.4° versus patients with no obstruction 9.7 ± 2.8°, p=0.02).17 In this regard, 3D STE provides a novel insight into LV mechanical alterations in select patient groups, such as those with HC.
The future of 3D STE appears bright. While the assessment of right ventricular (RV) function with 2D approaches has been limited to imaging planes, initial experiences18,19 with 3D STE have been promising, showing improved evaluation of right myocardial mechanics and understanding of RV function in patients with pulmonary hypertension. More data regarding normal values of myocardial mechanics obtained by 3D STE are needed.20–23 The results of an ongoing multicentre study focusing on normal values of 3D wall motion tracking in a variety of age groups (Normal values of 3D-wall motion tracking study) will be available in the near future.
Evaluation of Dyssynchrony
According to the American Society of Echocardiography Consensus Statement, ventricular dyssynchrony is a 3D phenomenon. Therefore, the assessment of LV dyssynchrony (LVD) is another important challenge and a promising application of newer techniques. This is particularly important considering that approximately 30 % of heart failure (HF) patients with left bundle branch block (LBBB) are classified as non-responders to cardiac resynchronisation therapy (CRT) – although progress has been made with the inclusion of magnetic resonance imaging in the evaluation process. Therefore, the recognition of myocardial activation delay by echocardiographic techniques is of clinical importance and has been a challenging area of research in past years. First experiences were based on the evaluation of timing intervals between segments derived by M-mode and pulsed Doppler techniques. Although this simple approach is helpful in individual patients, its limitations must be taken into account.
Tissue Doppler Imaging
Tissue Doppler imaging (TDI) was considered a promising technique, particularly for defining the segmental mechanical delay of systolic peak and time to onset or to peak velocity, and strain or strain rate values. Unfortunately, only small, single-centre studies with advanced expertise have been able to provide adequate results, thus limiting the reproducibility, broad acceptance and routine use of the technique. Time to peak velocity is often difficult to define and the velocity curves can be difficult to interpret. Advanced automated TDI techniques using different colour codes of segments with delayed mechanical activation were added to simplify the display of dyssynchrony. In this regard, angle-corrected TDI data for colour-coded temporal activation – a technique known as dyssynchrony imaging (Toshiba, Tokyo, Japan) – have the potential to provide radial dyssynchrony data that may be an additive to dyssynchrony analysis of longitudinal velocities. In 38 patients undergoing CRT, Dohi et al.24 demonstrated a sensitivity of 95 % and a specificity of 88 % of radial dyssynchrony for predicting an acute response to treatment. The data suggest that this technique is more robust than conventional velocity imaging.
Tissue Synchronisation Imaging
Tissue synchronisation imaging (TSI) (GE Vingmed, Horten, Norway) uses automated colour coding of time to peak velocities to define segments with delayed contraction, with red colour representing severe delay. This colour coding of temporal velocity data is superimposed on the routine 2D echocardiographic images to provide visual information about the anatomical regions of interest. Two studies comparing manual TDI versus automated TSI including a total of nearly 120 patients revealed adequate results (r=0.97 and 0.95, respectively, p<0.001).25,26 Furthermore, TSI studies in patients following CRT demonstrated that baseline dyssynchrony was significantly greater in responders than in non-responders, and correlated with volumetric change during follow-up.27,28 There are imitations of TSI, which include moving the region of interest within segments – possibly resulting in alteration of the measured delay – and incorrect timing – which may be a cause of error through inclusion of peaks outside the ejection phase. TSI as well as displacement and/or dyssynchrony imaging are TDI-techniques with known limitations based on the Doppler technology and the two-dimensional approach.
2D Speckle Tracking
2D speckle tracking has the advantage of differentiating between active and passive motion independently of Doppler angle of incidence and may be used to assess dyssynchrony.29–33 Whereas Becker et al.34 used CS to determine optimal lead position, others reported the value of longitudinal dyssynchrony as predicting response to CRT.35,36 A collaborative study (Speckle tracking and resynchronization [STAR]) between three centres found RS and transverse strain with a cut-off delay of 130 milliseconds between opposing walls to be predictive of LV ejection fraction (LVEF) response and outcome (death, heart transplant, LV assist device).37 However, since myocardial function occurs in three dimensions, a 3D approach should improve results and overcome some of the limitations of 2D speckle tracking.
3D Speckle Tracking
The initial clinical experiences and promising results in the evaluation of LVD using 3D STE were based on a pyramidal 3D data set and were demonstrated in 2008 by our institution.38 The advantage of the 3D STE technique is that it simultaneously calculates LVEF and end-diastolic and end-systolic volumes, and furthermore that it recognises segmental deformation related to time based on a 3D data set. We described that an alteration of the pacing mode after CRT (switching from the ‘off’ to the ‘on’ position) resulted in an immediate reduction of opposing segmental wall delay and end-systolic volume (see Figure 1). LVD based on 3D STE was estimated in all its components: LS, RS and CS, and the newly developed area strain (AS – see below) and global 3D strain. Tanaka et al. produced an early study on this technique, which focused on RS in 64 subjects.39 Strain measurement was feasible in the majority of patients studied. The authors found a close correlation between the 2D- and 3D-derived speckle tracking radial dyssynchrony parameters. In addition, the 3D site of the latest LV mechanical activation was successfully quantified in a manner previously considered impossible using 2D methods or conventional realtime 3D methods that focus on passive endocardial motion alone.40 Although both 2D and 3D STE demonstrated differences in sites of earliest activation in LBBB and RV-paced patients, the sites of latest activation were similarly distributed in predominantly posterior and lateral sites. A further study, also based on RS and dealing with RV pacing versus intrinsic LBBB using 2D and 3D STE, showed a more comprehensive LVD evaluation with 3D STE.41
Area Tracking/Area Strain
Area tracking/AS is a new modality based on 3D STE and reflects 3D strain and changes of the endocardium during LV contraction and relaxation. It is related to LS and CS and represents reciprocal values of RS. AS seems to be more robust and reproducible in comparison with other deformation parameters and may be a promising non-invasive tool to identify ischaemia. It may be possible to perform it in combination with stress echocardiography in order to improve detection and quantification of ischaemic response. It has further potential to localise delayed myocardial activation, which can be used in patients with LVD. Area tracking/AS was used for the first time by Thebault et al. in dyssynchronous patients.42 The AS-related standard deviation index reflects the dispersion of the 16-segment AS peak curves and allows a rapid and more global assessment of LVD in comparison with the use of a single parameter – for example, RS.Recently, quantification of mechanical LVD with narrow QRS duration has been reported by Tatsumi et al.43 The authors used the Area Strain Index (ASI), which was calculated as the average difference between peak and end-systolic AS of LV endocardium obtained by 3D STE based on the 16-segment model. The ASI score in HF patients with narrow QRS durations was lower than in those with wide QRS durations (2.5 ± 1.3 % versus 4.2 ± 1.2 %, p<0.001). Furthermore, the prevalence of high ASI scores in HF patients with narrow QRS was significantly higher than in normal controls, putting in question the results of the Resynchronization therapy in normal QRS (RETHIN Q) study based on conventional echocardiographic parameters.44 With the benefits gained by the addition of the third motion component,which remains invisible to both TDI and 2D STE techniques, this new technique may become the method of choice for the assessment of regional LV function.
Activation Imaging
Activation imaging technology (Toshiba, Tokyo, Japan) was introduced recently. It uses 3D STE, particularly 3D strain or AS, to define the onset of deformation in individual segments. As a colour-coded system, significant time delay is displayed in blue and pink colours according to the grading scale (see Figure 2). The algorithm uses a threshold for differentiating between varying physiological onset of deformation (see Figure 3) and distinct delay of mechanical activation (see Figures 4 and 5). This information can be displayed as a dynamic polar plot and bag image as well, and may be superimposed onto short-axis and long-axis planes (see Figure 3). Furthermore, distribution of delayed activation can be analysed frame by frame, providing a better overview of the extension of dyssynchrony. In a first series of consecutive patients treated with CRT in our institution, immediate change of delayed activation could be defined quickly and easily when the pacing mode was changed from CRT pacing mode ‘on’ to ‘off’ (see Figure 5).
Another potential of this technique could be demonstrated in individual patients with ischaemic heart disease. Patients with transmural myocardial infarction and regional wall motion abnormality may show mechanical delay related to the scar region. This phenomenon might be explained by a systolic sequence of primary stretch followed by recoil; however, these scar areas demonstrate distinct reduction in strain values. Interestingly, these data have only been reproduced in patients with subendocardial myocardial infarction and normal regional wall motion. Further studies are needed to evaluate the reproducibility and accuracy of this new technique.
Conclusion
The results of the Predictors of response to CRT (PROSPECT) study to define dyssynchrony raised questions about the potential and accuracy of conventional echocardiographic techniques, including M-mode, pulsed and tissue Doppler echocardiography.45 Newer techniques, such as 2D and 3D speckle tracking, have shown promising results in regard to outcomes and for defining the site and timing of delayed mechanical activation – with 3D STE having some advantages over 2D STE. Nevertheless, when using these newer techniques, we must face their specific limitations and the need for careful image tracing to manually fine-tune the regions of interest and capture the appropriate regional strain for dyssynchrony analysis. Inter-vendor variability of LV deformation measurements by 3D STE is another problem that will have to be solved.46 Newer developments, such as activation imaging based on 3D strain or 3D area tracking may further improve the potential of echocardiography to better select HF patients for CRT and follow their outcomes.