6.3 Interpretation
6.3.2 Speckle Tracking Imaging
155
156
method of analysis when using only the apical four chamber view as in this study. Individual segmental analysis would be the best method for analysis when using only the apical four chamber view. If territorial strain is to be used in the future, this study suggests that all three imaging planes (apical four chamber, apical two chamber and apical long axis) should be analysed. However, as explained earlier, the aim of this study was to ascertain trends within the data set. If there were significant correlations between the RCA and the presence and absence of RWMAs and CAD, these trends would show in the data. This, in general, was not the case.
It is surprising that the RCA region did not (in general) correlate well with RWMAs and underlying CAD as the inferoseptal wall was found to align with the transducer most of all the cardiac walls (reducing the angle of insonation) and there were never any significant artefacts in the region.
Moreover, a study published in 2009 by Gustafsson et al. [183] further confirmed the complexity of the intrinsic mechanics of the left ventricle. Using 40 healthy patients, they used circumferential ST to assess left ventricular motion. They found that the basal
inferoseptum segments rotated significantly more clock-wise than both the basal anterolateral segments and the segments at the mid and distal levels of the myocardium. They also noted significant differences concerning the duration of twist within the left ventricle (the basal segments duration of un-twisting was much longer than that of the apical segments). These differences in regional contraction play important roles in left ventricular filling and function.
The motion of the basal inferoseptum noted in the above study with circumferential ST may further explain the lack of statistical significance noted between ST and the presence/absence of RWMAs/CAD in the RCA segment in this CENSTEMI study. It is possible, that the accentuated motion of the basal inferoseptum counteracts a hypokinetic myocardium to some extent. Future studies may confirm this.
157
6.3.2.2 Limitations of ST Derived Velocity, Strain and Strain Rate
Although ST is a quicker and less observer dependent process than TVI [184], there are various pitfalls to this technique which could contribute to the varied correlations and overlap noted in this study. ST relies upon a high frame rate (50-70 fps [99]) which can be difficult in individuals with borderline image quality.
The author found frame rates were often difficult to perfect, due to the reduced quality of patients individual images (frame rates ranged from 36 to 109 frames per second). An
increased frame rate reduces under-sampling, however, results in decreased spatial resolution (the ability of a system to distinguish two points as separate in space), which results in less than optimal tracking of the region of interest [185]. Decreased frame rates, increase spatial resolution. But, because the software searches and tracks individual speckles frame-by-frame, this may mean that, owing to the low frame rate, a speckle is outside of the search area, once again resulting in less than optimal tracking of the region of interest [181]. There is a chance, therefore, that variations in frame rates may contribute to underestimated or overestimated values. However, Edvasden et al. [89] used frame rates ranging from 56-134 fps with excellent results, furthermore, studies have suggested that frame rates >30fps allow for reliable and reproducible software operation [186, 187]. This means that the frame rates used in this CENSTEMI study were acceptable.
ST is a two dimensional measurement, whereas cardiac motion is three dimensional [99].
This means that the modality may not capture the full motion of the left ventricle. Therefore, areas may be under or over estimated. Values are variable with age and sex, so there is a possibility that a normal value for one patient may be borderline or abnormal for another [188, 189]. This study was looking for trends in a predominantly older subset of
predominantly male patients, so these factors, although still relevant, should be minimised.
158
6.3.2.3 Statistical Significance between the LAD Territory and ST Derived Velocity Strain and Strain Rate
The strain and strain rate values for the LAD region correlated best (were statistically significant) with the presence and absence of RWMAs and underlying CAD for both the TVI and ST modalities. This may be due to a large number of LAD infarcts in this research (15 patients had LAD single vessel CAD, compared with 5 patients with RCA single vessel CAD and no patients with Cx single vessel CAD). It may also be explained by less individual variation in the LAD coronary artery distribution [190]. Good cursor alignment and
visualisation of LAD regional segments may also have assisted with the favourable results.
6.3.2.4 Benefits of the Utilising of Speckle Tracking over Tissue Velocity Imaging
Unlike TVI, which requires user to determine sample volume positioning and waveform measurement, ST requires only a user defined region of interest which is optimised to follow the endocardium throughout one cardiac cycle. The result is automated analysis of segments which are less influenced by the user. Although, an average of three measurements is recommended to ensure reproducibility.
This CENSTEMI study found ST much quicker to perform, which agrees with literature in the field [105, 191]. When the region of interest was optimised, automated analysis resulted in velocity, strain and strain rate waveforms. Values for peak velocity, strain and strain rate were analysed, pre-measured and displayed automatically in a box below the waveforms. Although a high frame rate was required for ST analysis, only one cardiac cycle was required. The need for just one cardiac cycle meant that the traces were less likely to contain incidental movement or inspiration. Hoyt et al [103] suggested ST to be less time consuming as it is a semi-automated process, beneficial over TVI as studies suggest increased reproducibility, increased signal to noise ration and no angle dependency.
159 6.3.2.5 ST Summary
This study shows that ST of the apical four chamber view correlated well with strain and strain rate with the presence of both RWMAs and underlying CAD in the LAD region.
However, less correlation was noted in the Cx region, and lesser still in the RCA region.
This may be due to limitations with the study methodology such as theoretical coronary artery distribution assumptions, difficult patient image quality, the complex mechanics of the left ventricle, or perhaps reflects the real-world application of ST.
160