Background
Introduction of tissue velocity echocardiography (TVE) two decades ago[
1] has opened new possibilities for non-invasive quantification of myocardial function. TVE methodology relies on the detection of the low velocity/high amplitude motion of myocardial tissue by application of appropriate low pass filtering to the received Doppler signal in order to distinguish it from the high velocity/low amplitude motion of the blood [
2,
3]. The diagnostic performance of TVE has been extensively studied and its incremental value for the evaluation of hypertrophic cardiomyopathy [
4], coronary artery disease [
5,
6], and the determination of systolic and diastolic left ventricular function [
7,
8] is well documented.
Tissue Doppler in pulsed mode (pulsed/spectral TD) [
9,
10] and colour TD [
11,
12] are the two modalities available today in tissue velocity echocardiography (TVE). Spectral TD registers the instantaneous frequency spectrum at the chosen myocardial region and the signal is computed using Fast Fourier Transform (FFT) technique in order to obtain the myocardial velocity distribution. On the other hand, colour TD relies on autocorrelation analysis to provide average Doppler frequencies for a chosen set of pixels and thus yields mean myocardial velocities in the interrogated myocardial region. Both TVE modalities have been validated in experimental [
13,
14] and clinical studies [
9,
11,
15] and the requirements for optimal signal sampling [
16] as well as the effect of instrumentation settings on the accuracy of TD measurements [
17,
18] have been evaluated.
However, despite the fact that the two tissue Doppler modalities target the same physical quantity i.e. velocity, they operate in different ways and, indeed, a number of comparative studies performed hitherto have shown poor agreement between the results obtained with the two modalities [
17,
19,
20]. This is not surprising, since the general applied method of myocardial velocity estimation with spectral TD involves measurements performed at the outer border of the spectral velocity wave. Consequently, this procedure identifies maximal components of the velocity spectrum in the region of interest, whereas with colour TD modality, a mean velocity at the interrogated myocardial location is obtained. Hence, it appears reasonable to assume that spectral TD velocity measurements based on averaged, rather than maximal signal, would better agree with the results of corresponding measurements obtained with colour TD technique. Since, to our knowledge, this issue has not yet been specifically addressed as yet, the aim of this study was to compare the results of myocardial velocity and displacement measurements using colour TD with those obtained with spectral TD based on averaging of the spectral signal.
Discussion
The present results demonstrate that longitudinal systolic myocardial velocities derived from colour TD are not only closely correlated but also in good agreement with mean spectral TD velocities. A similar degree of concordance and correlation was observed for LV systolic myocardial displacement measured by colour TD and mean spectral TD procedure. This is not entirely unexpected when considering the physical principles of the two TVE techniques. Indeed, regarding spectral TD, the acquired composite frequency signal is subjected to Fast Fourier analysis and the sum of velocity distributions from the individual scatterers passing through the range gate is obtained at time intervals defined by the pulse repetition frequency. The spectral signal reflects thus the sum of velocity distributions at any instant during the recording period. Assuming Gaussian distribution of instantaneous myocardial velocities at the interrogated region [
24], mean velocities are to be found in the mid portion of the spectral signal while the outer and inner contours of the envelope reflect maximal and minimal myocardial velocities, respectively. Colour TD, on the other hand, relies on autocorrelation algorithm in order to estimate the average frequency shift for a given set of pixels along the Doppler sector and provides consequently mean myocardial velocities at the region of interest. Hence, it can be expected that colour TD and mean spectral TD velocities would closely relate to each other.
To our knowledge, there are no studies published hitherto that compare directly colour and mean spectral TD velocities. Recently, Chen et al. have reported that systolic as well as early diastolic longitudinal velocities measured at the mid portion of the spectral signal were nearly equivalent to those obtained by digitized M-mode recordings at septal and lateral site of mitral annulus [
22]. Our findings add to these observations and demonstrate that colour TD and mean spectral TD procedure provide almost identical values when measuring systolic velocities. However, the results of myocardial velocity measurements with colour TD and mean spectral TD procedure may vary between -2.03 and 1.85 cm/sec (lateral wall), a fact that should be kept in mind in the clinical setting, especially when low longitudinal velocities are recorded as in the case of ischemic myocardium. The slight non-significant numerical underestimation of mean spectral velocities by colour TD can, at least partly, be explained by the higher inherent temporal resolution of pulsed TD. Despite the unusually high frame rate employed in this study for colour TD recordings (>250 Hz) and the fact that measurements were performed on unfiltered data, the temporal resolution obtained was still well below that of spectral TD recordings and this might result in a possible underestimation of minor myocardial motions. Furthermore, Walker et al. demonstrated that spectral TD tended to overestimate tissue velocities [
25], an observation well in accord with the present findings.
The current results reveal that colour TD and mean spectral TD myocardial velocities are concordant even when temporally integrated. While the employed software provides a computerized temporal integration of colour TD velocities in order to assess myocardial displacement, in the case of spectral TD, manual tracing of velocity spectra is required for estimation of myocardial motion. Nevertheless, no statistical differences between the displacement measurements performed with colour TD and mean spectral TD procedure appeared between the two interrogated myocardial walls. Still, our results imply that displacement values measured by the two TVE modalities may differ as much as 4 mm (lateral wall), which should be considered in the clinical situation.
Furthermore, similar to what was observed in the study of Lind et al. [
26], the present M-mode measurements of mitral annulus motion at the lateral wall produced displacement values that were significantly higher than those obtained with colour TD and mean spectral TD method. The overestimation of colour TD displacement by M-mode on the lateral wall was recently confirmed by Ballo et al. [
27] and the authors reported that the agreement between M-mode and spectral TD measurements was improved when adjusted (mean) spectral TD signal was used. The present results are in keeping with these observations even if the discrepancy between M-mode and colour TD measurements in the present study was less pronounced, possibly as a result of the higher colour TD sampling frequency currently used.
The overestimation of both colour and mean spectral TD displacement at the lateral wall site observed by M-mode may be partly the result of an unfavourable angle of incidence. Indeed, an optimal insonation angle is difficult to be achieved at the lateral wall site. For TVE measurements, deviation of the incidence angle would lead to underestimation of longitudinal myocardial velocities and displacement and the underestimation would be directly dependent on the cosine of divergence angle of the Doppler beam from the longitudinal plane. On the other hand, an unfavourable insonation angle of the M-mode beam would result in overestimation of the true longitudinal motion due to the influence of concomitant radial myocardial movement as the mean vector of radial and longitudinal motion is measured.
The lower interobserver variability for current spectral measurements compared with what was previously reported [
15,
28] reflects most probably standardization of the present TD recordings and exclusion of signals with uneven spectral envelope contour. However, it still reflects the subjectivity in identifying the borders of spectral envelope. The reproducibility of the present colour TD velocity measurements was better than that of spectral TD and similar to that previously documented in a larger study [
29]. Furthermore, in order to approximate the high inherent repetition frequency of the spectral TD and accurately register minor myocardial motions, colour TD recordings were performed employing high frame rates without applying temporal filtering on off-line measurements [
16,
18]. However, increased temporal resolution may as well yield an increased noise to signal ratio which in turn may influence the results obtained with colour TD analysis. The interobserver variability values obtained in the present study was similar to those reported previously [
29] and any significant confounding influence of high sampling frequency appears thus to be unlikely. Finally, the number of individuals included in this study was rather small and limited on healthy volunteers. Larger studies on consecutive groups of patients are needed in order to conclusively confirm the present results and the interchangeability of colour TD and spectral TD procedure.
Up to date, specific technical and measuring recommendations for spectral measurements are lacking. It is generally recommended that gain settings should be adjusted at a minimal level in order to avoid spectral broadening but yet the technical requirements are not specified which may, at least partly, explain the discrepancy observed between different TD studies. Indeed, while significant correlation between systolic myocardial velocities measured with pulsed TD and left ventricular ejection fraction has been found in a number of studies [
30,
31], other reports based on colour TD show only weak association between these two indices [
29]. In the present study, spectral recordings were performed applying transducer settings conventionally used as default by the manufacturer with transmit gain of 0.0 dB, receive gain of 3.0 dB and off-line gain saturation 50% saturation.
Regardless the aforementioned limitations, the results of the present study have some important clinical implications. Chen et al. in their study have shown that mid envelope (mean) spectral velocities were almost identical with velocities measured with digitized M-mode. The authors concluded that their findings together with the lack of standardized measuring conventions for spectral TD recordings could imply that mean spectral TD procedure may be recommended in clinical routine [
22]. However, the application of mean spectral TD method, especially for displacement measurements, is rather laborious and time consuming which may jeopardize the use of this method in clinical praxis. On the other hand, based on the present results showing that colour TD provides velocity and displacement measurements that are in good agreement with mean spectral TD and M-mode calculations, the use of colour TD with sufficiently high temporal resolution may be recommended as an efficient and reliable tool for quantification of myocardial function in clinical situation.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AR contributed to the design of the study, performed measurements and calculations from ultrasound data as well as statistical analysis, and participated in the interpretation of the results and preparation of the manuscript. AS performed measurements and calculations from ultrasound data as well as statistical analysis. EN participated in the interpretation of the results and preparation of the manuscript. PV contributed to the interpretation of the obtained results. KS participated in data collection and interpretation of the results. RW and LÅB supervised the study and contributed to the interpretation of the obtained results. JN supervised the study, contributed to the analysis and interpretation of the data, and was responsible for the preparation of the final version of the manuscript. All authors read and approved the final manuscript.