The currently observed differences between the results provided by the three different echocardiographic modalities is not entirely surprising since the methods represent three independent approaches to the assessment of systolic LV displacement. Both spectral and colour TD method provides quantification of regional myocardial velocities as a base for a subsequent calculation of myocardial displacement. However, different methodologies are used for the computation of velocity curves. Spectral TD detects frequency shift between emitted and returning ultrasound beam and the acquired data are then analysed using Fast Fourier Transform technique . The obtained velocity wave contains a spectrum of velocities from the interrogated myocardial location and the maximal velocity can be measured at the envelope of the velocity curve. Different from spectral TD, colour TD estimates frequency shift indirectly using autocorrelation analysis and provides instead average of all velocities within the myocardial region of interest . Hence, the maximal velocity values obtained with colour TD should be expected to be lower than those extracted from spectral TD velocity wave.
In contrast to spectral and colour TD, anatomic M-mode focuses on the direct measurement of the amplitude of the myocardial motion during cardiac cycle by analysis of 2D samples in time domain along freely positioned virtual M-mode line and the information is then displayed in a way similar to that of conventional M-mode [14, 15]. Measurement of LV dimensions and atrioventricular plane movements during systole using conventional M-mode has been found to be an accurate method for the assessment of regional and global left ventricular function [9–11, 26] The anatomic M-mode technique further increases accuracy of the M-mode measurements by reducing possible errors produced by inappropriate alignment of M-mode ultrasound beam non-orthogonal to the interrogated myocardial segment . Overestimation of the systolic LV shortening, especially in the lateral atrioventricular margin where the measured displacement would represent a mean vector of the longitudinal and the concomitant radial LV motion, can thus be avoided. The accuracy of anatomic M-mode has been found to increase with second harmonics imaging, but to decrease at correction angles > 60° . Therefore, in order to ensure optimal accuracy of the performed measurements, all images in the present study were acquired in second harmonics mode, and the correction angle never exceeded 30°.
When considering the results of the current comparison between the three echocardiographic modalities, it should be kept in mind that for technical reasons, the measurements with spectral TD could not be performed on the same cardiac cycles as those chosen for the colour TD and colour-coded anatomic M-mode measurements. Although the spectral and colour TD imaging followed each other closely and no significant heart rate alterations occurred between the respective registrations, a possible influence of the separate image acquisition on the relationship between the measured variables still cannot be excluded. In addition, both spectral and colour TD measurements are prone to be affected by alterations of insonation angle and the influence of such variations on the obtained results cannot be entirely excluded either. However, all image acquisition in the present study was performed by the same experienced sonographer and a special care was taken to keep the Doppler angle parallel with the LV long axis, with the sampling volume at the same location within the myocardium. Such a procedure would certainly minimize any possible effects caused by random alterations of insonation angle. Furthermore, it appears unlikely that any of the above-mentioned confounding factors would act exclusively in one direction, and the internal consistency of the present results showing clear differences between the evaluated modalities contest any significant blurring effect of these factors.
The present results reveal the occurrence of significant differences between the results of displacement measurements obtained with the three employed echocardiographic methods. Although the mean (all four LV walls) displacement values obtained with colour TD was statistically significantly higher than the corresponding values produced by anatomic M-mode, the magnitude of the difference seemed to be clinically negligible. The constant difference between the anatomic M-mode and colour TD mean displacement values was only -0.2 ± 0.8 mm. However, these figures also imply that the displacement measurements by the two methods may differ by as much as -1.8 to +1.4 mm. In view of the current interobserver variability of anatomic M-mode (7.4%) and colour TD (6.9%) that would produce variability of the respective results by not more than ± 0.8 mm, these limits of agreement, even if fairly narrow, can hardly be clinically accepted. In addition, the limits of agreement for the anatomic M-mode and colour TD measurements in individual LV walls were even wider [see Additional file 3].
The current results are in keeping with those reported from the study of Ballo et al.  in which spectral TD overestimated the displacement values obtained with the conventional M-mode method employed by these authors to the same degree as it was found in the present experiments. At the same time, the authors reported greater than that currently observed overestimation by M-mode of the results produced by colour TD method. However, the measurements in the above-mentioned study  were performed exclusively in lateral LV wall and it is often difficult to achieve an appropriate angle of incidence in this LV region. Consequently, the magnitude of the observed difference between conventional M-mode and colour TD modalities might to some extent reflect possible angle-dependent overestimation of the true LV ejectional shortening by M-mode. Similar results were reported from the study of Yumi Hayashi et al. , in which the colour TD displacement was generally slightly lower than the displacement measured with anatomic M-mode. The small discrepancies between the present results and those reported by Yumi Hayashi et al.  may be caused by the fact that, contrary to the present measurements, the displacement values in the above-mentioned study also included isovolumic myocardial movements.
The currently observed overestimation of the results of anatomic M-mode and colour TD measurements by spectral TD was not only statistically but also clinically significant. The present results are in keeping with the previous reports of significantly higher systolic velocity values measured by spectral TD than by colour TD modality [18, 19] and demonstrate that spectral TD displacement values are higher than those provided by direct displacement measurements with anatomic M-mode as well. In addition, reproducibility of spectral TD measurements of myocardial motion velocities has been demonstrated to be low , and even if the process of temporal integration of velocity curves for calculation of myocardial displacement has by itself a smoothing effect, both the dispersion of the current individual spectral TD displacement values as well as the interobserver variability was clearly higher than the respective values for the two other methods. Consequently, the agreement between spectral TD and these methods was poorer than what was observed with colour TD and anatomic M-mode. The constant differences between spectral TD and the two other modalities amounting to almost -6 mm and the wide limits of agreement with anatomic M-mode (at least -9.6 to -2.0 mm or wider) and colour TD (at least -8.4 to -1.2 mm or wider) cannot be clinically accepted and discourage the use of spectral TD for the measurements of myocardial displacement interchangeably with the two other methods.
The present study demonstrates that beside the choice of measuring modality, also adjustments made by sonographer during echocardiographic data acquisition and analysis can significantly influence the accuracy of the myocardial displacement measurements. In fact, Lui et al. identified the Doppler gain as one of the most significant sources of error and variability in in vitro model of pulsatile flow . Increasing gain would result in increasing spectral broadening and upward shift of the upper border of the spectral velocity curve, with higher measured velocity values as a consequence. Not surprisingly then, the effect of offline gain saturation on the results of the present spectral TD measurements was indeed highly significant. Furthermore, it is important to emphasize in this context that the magnitude of the currently observed gain saturation effect on the spectral TD displacement, both in terms of measured values and their dispersion, is of crucial clinical relevance since the measurements with unsaturated and fully saturated offline gain lacked any clinically acceptable agreement at all [see Additional file 4]. In contrast, the effect of temporal filtering on the present colour TD measurements was barely measurable and clinically negligible, thus rendering this modality more robust in the clinical setting.
Measurement of LV myocardial longitudinal motion with spectral and colour tissue Doppler technique is now a generally accepted procedure for the assessment of global and regional LV function . Together with the conventional M-mode still broadly used for this reason, the two Doppler based methodological approaches have been hitherto often used interchangeably. However, these technologies have entered clinical diagnostic laboratories long before their impact was fully recognised and there are still only limited data on their effectiveness in population screening in the high volume echocardiographic laboratories. In fact, the effectiveness of colour TD in the detection of coronary artery disease has been addressed in some large-scale studies [31, 32] but similar studies addressing spectral TD remain still to be performed. Against this background, the present results have important clinical implications since they reveal the occurrence of systematic differences between the results produced by different tissue Doppler and M-mode based methods and highlight a need of reassessment of the diagnostic qualities of the Doppler modalities. The currently observed differences between the three evaluated echocardiographic modalities and the different degree of sensitivity of the tissue Doppler methods to offline processing do not support their interchangeable use and emphasize the need of specific for each method normal values when comparing spectral, colour and anatomic M-mode measurements. Furthermore, the observed high sensitivity of spectral TD to offline gain setting calls for strict standardization of the offline data processing when using this method, or alternatively, the use of gain level specific normal values. This, together with the recently reported low reproducibility of spectral TD velocity measurements raises, however, serious doubts about the applicability of this modality in the routine echocardiographic practice.