Abstract Background Speckle tracking based global longitudinal strain (GLS) values have proven useful in the assessment of subtle changes in left ventricular function. From a clinical point of view, robustness and reliability of measured values are critical to ensure a valid patient assessment and follow-up. However, it is still a matter of debate if imaging parameters systematically alter measured strain values and if these changes are relevant as compared with GLS fluctuations that are caused by different operators or different studies by the same operator. Methods In a consecutive everyday patient population (n = 35), we recorded the apical four chamber view several times in each patient with different ultrasound machine settings (modification of gain, frame rate, sector depth, and transducer frequencies) using a commercially available ultrasound imaging system. Furthermore, apical four chamber views with ‘optimized’ imaging settings at the operators’ discretion were recorded by two different observers (obA/obB) in each subject to compute inter- and intra-observer variability. GLS values were calculated offline with a dedicated software. We fitted a linear mixed effects model with random intercept and slope to assess the effect of imaging parameters on GLS and compared the two investigators with Bland-Altman plots. Results Ejection fraction ranged between 10% and 76% and was correlated well with GLS (r = -0.78). Neither gain settings (range: -24 to 24 arbitrary units, p = 0.68) nor frame rate (range: 51-113 sec-1, p = 0.77) systematically changed measured GLS values. Conversely, higher sector depth increased (range: 12 to 24 cm, mean effect: -0.16%/cm; 95% CI -0.24% to -0.07%, p < 0.01), and higher transducer imaging frequencies decreased absolute GLS values (range for harmonic imaging: 1.5/3.1 MHz to 2.0/4.3 MHz, mean effect 1.10%/MHz; 95% CI 0.61% to 1.59%, p < 0.01). According to our data, a 12 cm increase of sector depth would translate into an average change of -1.87% GLS (95% CI: -2.87% to -0.86%), whereas a switch of the second harmonic imaging frequency from 3.1 MHz to 4.3 MHz would cause a 1.32% GLS change (95% CI: 0.73% to 1.91%). Intra- and inter-observer variability showed good correlation and limits of agreement (obA: mean difference [MD]: -0.20%; 95% limits of agreement [LOA]: -2.42% to 2.02%, p = 0.86, obB: MD: -0.10%; 95% LOA: -4.28% to 4.07%, p = 0.12, obA vs. obB: MD: -0.53%; 95% LOA: -3.68% to 2.62%, p = 0.92). Conclusion Overall, GLS values were robust and reproducible in our cohort of patients. In comparison, potential systematic changes of GLS values caused by modification of imaging parameters (sector depth/transducer frequency) were much less in number than GLS variations caused by different operators or different studies by the same operator.
Read full abstract