Author + information
- Benjamin C.F. Smith, MSc, DIC∗ (, )
- Gary Dobson, MDCM, MSc, DIC and
- Petros Nihoyannopoulos, MD
- ↵∗Echocardiography, Hammersmith Hospital, Du Cane Road, London W120HS, United Kingdom
We thank Prof. Flachskampf for his interest and comments regarding our recent paper (1).
The full characterization of the 3-dimensional (3D) strain tensor requires its instantaneous magnitude and direction and is decomposed into 3 tensile strains and 3 shear strains. Assuming a constant ventricular mass, area strain (AS) is a function of all 3 tensile strains because endocardial area can only decrease if radial length increases and thus is less sensitive to directional changes. This is illustrated in Figure 1, where directional change is effected by multiplying radial strain (RS) by 0.8 through 1.15. Circumferential strains (CS) and longitudinal strains (LS) vary from baseline by a factor of 0 to 2, whereas AS varies by a factor of 0.85 to 1.1. Furthermore, AS did not vary for a given RS, in contrast to CS and LS. The statistical explanation for AS bumping the derived strains out of the logistic regression is therefore clear; the individual tensile strains can add no further information to AS.
The predicted arithmetic relationship between AS, CS, and LS will be correct on average. However, as with all models of biological behavior, there may be considerable segmental variation within and between individual subjects. This variation would be amplified by attempts to calculate AS from CS and LS derived from 2-dimensional (2D) speckle tracking, where there would be doubt regarding concordance of measurement sites.
When using a perceived strength of one methodology to critique another, one must take into consideration the limitations of the former. There is a saying: “if you see the world in black and white, you’re missing important gray matter,” and so it goes with 2D strain. The missing gray matter in this case are speckles that are “lost” as they leave the thin 2D slice plane as the heart twists obliquely. To compensate for this speckle loss, the program must constantly find new speckles and correlate their movement to the speckles that were originally tracked. As such, temporal resolution is quite important for 2D strain. 3D strain, on the other hand, has much less speckle loss because the data remain within the zone of interrogation and therefore the frame rate may be less crucial.
The fact that speckles are able to be followed in 3D allows the program to follow true myocardial movement. Comparing 3D movement with a single plane analysis would be like comparing apples and oranges.
For a new technique to be valid, it is important to assess its reproducibility and link it to clinical outcomes; we have successfully done that (1). The optimal frame rate for speckle tracking clearly remains a matter of debate and may remain so until a time when new machines are released with improved frame rates. At such time, the perceived temporal limitation of 3D strain may become irrelevant.
- American College of Cardiology Foundation