Introduction Echocardiographic strain measurements can be performed with 2D or 3D speckle tracking. Differences between 2D and 3D measurements are reported, but the exact mechanism for this is unclear [1]. Also, measured myocardial strain magnitudes in fetal hearts vary substantially across different studies [2], and it is unclear why this is so. Our study aims to demonstrate possible reasons for these discrepancies. Methods 4D echocardiography images (STIC mode) were obtained from 10 heathy fetuses. A validated cardiac motion estimation algorithm [3] was used to track the motion of the fetal left ventricle (LV) in 3D and 2D, and was used to calculate myocardial strains. The same images were used for both 2D and 3D strain quantification to enable controlled comparisons. Results 3D longitudinal strain (LS) was significantly lower than 2D LS by 2.7%. This is partially due to LV twist causing the longitudinal line in 3D to move out of plane while preserving its length after contraction. 2D quantification cannot capture this effect. By quantifying strains in a way to negate LV twisting (tracking motion in 3D but projected tracked points to the 2D plane before strain calculation), We found that 1.2% of the difference can be explained by LV twist motion. In the circumferential direction, 3D strain was found to be significantly higher than 2D circumferential strain (CS) by 2.0%. This can be fully explained by the systolic motion of myocardium towards the apex, which brings wider transverse cross-sections down to the imaging plane. A timing mismatch was observed between when the longitudinal and circumferential lengths are at their peaks, caused by LV shape changes during the isovolumic contraction period. In 2D strain quantification, strain in each direction is assigned different zero-strain reference time; in 3D, a single reference time is used for both directions. Favouring any one direction when specifying this reference will reduce strain magnitude of the other direction. This accounted for another 0.7-0.8% difference between 2D and 3D strains. A spatial variability of strains was also found. Strains at epi- and endo-cardial locations differed substantially, by 3.6% in the longitudinal direction and 9.3% in the circumferential direction. Since strain quantifications are manually controlled clinically, this could account for wide discrepancies between the different studies [2], in which reputable groups and publications reported strain values that differed by 6.3-7.1%. Different smoothing extent during motion tracking causes significant differences in strain values, and could be another factor for discrepancies between studies. Lastly, we find that 3D motion tracking can enable the quantification of LV twist (7.8±3.3°) and the average myofiber orientations (4.6±2.7°). Myofiber orientations was estimated via eigenvectors of the strain tensor. Discussion We demonstrated mechanisms for discrepancies between 2D and 3D strain measurements, and potential reasons for wide discrepancy in literature values of fetal myocardial strains. Our finding calls for caution in interpreting strain results in the literature before future standardization of strain is achieved. Although our study was conducted on fetal echocardiography, the results are likely applicable to adult echocardiography.
Physiology 2021 (2021) Proc Physiol Soc 48, PC090
Poster Communications: Factors Contributing to the Discrepancies In 2D and 3D Fetal Echocardiography Strain Measurements
Meifeng Ren1, Wei Xuan Chan1, Aimee Armstrong 2, Andreas Tulzer3, Gerald Tulzer3, Martin Lindsay Buist1, Choon Hwai Yap4
1 Department of Biomedical Engineering, National University of Singapore, Singapore, Singapore 2 The Heart Center, Nationwide Children’s Hospital, Columbus, ohio, The United States of America 3 Dept of Pediatric Cardiology, Kepler University Hospital, Linz, Austria 4 Dept of Bioengineering, Imperial College London, London, United Kingdom
View other abstracts by:
Where applicable, experiments conform with Society ethical requirements.