,
27 tweets,
23 min read
Read on Twitter
For all those unable to make my 7 AM presentation on Speckle Science yesterday, here's the requested #Tweetorial on Strain basics #ASE2019 @ASE2019 as promised.
Myocardial fibers make an orderly transition from a L handed helix in the subepicardium to a R handed helix in the subendocardium with circumferential fibers in the mid myocardium. 
This creates 3 fundamental motions: radial thickening toward the centroid, circumferential shortening/twisting, and longitudinal shortening.
Strain is a property from mechanics: deformation caused by application of a force. Myocardial strain is a % or proportion. Strain can be Eularian (sum of instantaneous change in length) or Lagrangian (sum of changes over initial length).
You can think of this like a balloon - if you place two dots on a balloon at an initial length (L0) and expand it, the dots will move further apart (L). Thus, lengthening is a/w positive strain and shortening with negative.
LVEF is dependent on EDV and thus may reflect remodeling rather than contractility. It has limited reproducibility and is a global measure (and not as sensitive to regional changes). Strain has robust outcomes data to identify preclinical changes in altered contractility.
Strain can be measured by #EchoFirst or #whyCMR (the latter by applying tags to the myocardium and tracking over time vs tracking blocks of tissue)
Speckles are a result of ultrasound interference (not from actual anatomic structures) and result in a fingerprint that can be tracked over time with texture and contour.
Strain calculation starts with identifying end diastole (end of R wave) and end systole (AVC or smallest LV volume). Subsequently, semiautomated borders are applied to define the region of interest, these are segmented, and those segments are tracked.
Segmentation separates the myocardium into blocks of pixels. The algorithm then identifies how these blocks of pixels move by attempting to pixels that appear the most similar to the initial block and calculating vectors of motion.
Once a pattern is matched (even if not identical in shape or size), changes in length are determined.
Vendors have unique algorithms to reduce noise through temporal and spatial smoothing, though these may mask real findings.
Gain is important (don't want to be over or undermined) but strain values can still be calculated.
Lots of validation against phantoms, clinical models (with animals) and in humans. In general, longitudinal strain is more reliable than circumferential or radial and global strain better than regional.
#3D strain can be thought of as a % change in area (as the endocardial surface shortens during systole) as opposed to a change in length.
This can be segmented into transverse, sagittal, and coronal sections.
This can identify regional and global aberrations in contraction and rotation.
