LAUSR

The idea of qualifying motion of the myocardium as a sign of its pathologic function has been around already since early MRI. In 1988, Zerhouni et al proposed to label myocardium and follow it during contraction. This has been further developed to give all the specialized MRI tracking sequences: Spatial Modulation of Magnetization (SPAMM), Displacement Encoding with Stimulated Echoes (DENSE), and Strain-Encoded (SENC) imaging. The application of feature tracking algorithms, designed originally for echocardiographic image analysis, to cardiac magnetic resonance cine images is a novel approach with potential advantages over existing methods. Such a technique would eliminate the need for additional time-consuming scans designed specifically for strain measurement. Additionally, the described method of cardiac magnetic resonance feature tracking is quite rapid and mostly automated. As suggested by Amzulescu et al, myocardial tissue tracking imaging techniques have the potential to contribute, incremental to ejection fraction, to the diagnosis of cardiac diseases. cvi42 tissue tracking software (Circle Cardiovascular Imaging, Calgary, Canada) is supposed to construct a deformable myocardial model based on the tracking, assuming the myocardium is nearly incompressible. It has been validated against other tracking software. Since in scar tissue healthy myocardial muscle is replaced by fibrosis, it seems obvious that dyskinetic behavior should be expected in cardiac segments with scar, but as pointed out by Alessandro Pingitore et al there is no one-to-one matching between mechanical markers and scar, akinetic myocardium without scar may exist in some pathologies. It has been underlined for a long time in different reviews, such as the one from Chitiboi and Axel that while global function measures like ejection fraction are commonly used in the clinic, MRI also provides a range of approaches for quantitatively characterizing regional cardiac function, including the local deformation, but they have had relatively little clinical evaluation. In this issue of JMRI, an article by Zhao et al advances our understanding of local strain quantification by reporting, in 109 patients with ischemic heart disease, global and segmental level data for longitudinal and circumferential strain. This quantification has been done with strain taken from the tissue displacement plugin of cvi42 commercial software. These values come from 109 patients and 50 volunteers carefully and retrospectively selected in a population of 179 patients and another population of healthy subjects. In clinical praxis by looking at rest cine MRI, even without feature tracking, clinicians already report dyskinesis, but qualitatively. The availability of the feature tracking software analysis with easy contour propagation is one piece that has enabled production in such a large population of quantitative measurements. A quite significative portion (17%) of the recruited potential population had to be excluded due to insufficient image quality. Since ischemic heart disease may not be the only pathology presenting motion impairment, the authors chose the population ruling out other causes, like heart failure. All ischemic heart pathologies are well documented with proven coronary artery stenosis, stress and rest perfusion cardiac magnetic resonance examination, and late gadolinium-enhanced MRI. The patient population and the segments are independently divided into three categories: tissue presenting late gadolinium enhancement, tissue without enhancement but ischemia perfusion stress/rest mismatch, and tissue without ischemia nor enhancement. The use of a Z-score to account for the variability of strain observed in different segments, even in a healthy population, is an elegant analysis tool that enables to group data coming from all segments for a category. Differences in longitudinal and circumferential strain between the four groups, three pathologic categories and healthy, are analyzed and statistically significant differences are observed between groups. Differences between groups are of less than 2% and observed variances within groups are on the order of 2–5%.