$\textbf{INTRODUCTION:}$ Myocarditis a rare hearth disease, more common with young adults. As an alternative diagnostic method, speckle tracking echocardigraphy (STE) is used, which is more accessible as other established methods. Changes in global strain can be detected with this method, while location and extent of inflammation remain unknown due to big variability of measurements. In this paper, we researched the causes of variability influencing the quantitative evaluation of ultrasound measurements.
$\textbf{METHODS:}$ Longitudinal strain curves of the left ventricle were measured with STE in 34 patients upon the diagnosis and 6 months later, in which we detemined the point of maximum deformation $\varepsilon_P$. With the analysis of average $\varepsilon_P$ in sectors, projections and slices, and the calculation of its gradient from base to apex (k) we determined the effects on variability: three different projections, deviating measurements, method of calculation of $\varepsilon_P$ and interpersonal differences. We assessed these effects by calculating mean and standard deviation.
$\textbf{RESULTS:}$ Average calculated $\varepsilon_P$ was –20.3 $\% \pm$ 5.2 $\%$. Average $\sigma_P$ in projections was 5.2 $\%$, k –2.15 $\% \pm$ 2.68 $\%$, with no statistical differences (p>0.62 and 0.27). The effect of the deviating measurements was in the expected range. When calculating $\varepsilon_P$ from multiple points, the average $\sigma_P$ was 5.1 $\%$, k –2.04 $\% \pm$ 2.66 $\%$ in projections, with no statistical difference from previous values (p>0.26). Interindividual variability of healty individuals and patients was 30 $\%$ higher than intraindividual variability before and after tratment.
$\textbf{CONCLUSION:}$ Local properties reflect the known mechanics of the left ventricle. Most of the variability of $\varepsilon_P$ stems from the natural interindividual variability of the deformation since the optimization of the quantitative analysis (selection of measurement projections, exclusion of deviating measurements, improvement of the calculation of $\varepsilon_P$) does not contribute significantly to its reduction.
|