AUTHORS: Najdenovska E, Masci PG, Giulia Pavon A, Bach Cuadra M, Schwitter J

Society for Cardiovascular Magnetic Resonance (SCMR) Annual Scientific Sessions, : , Seattle, USA, February 2019

ABSTRACT

Background:

In patients with acute ST-elevation myocardial infarction (STEMI), microvascular obstruction (MVO) is generally accepted as one of the strongest predictors of adverse outcomes. However, quantification of MVO mass is not trivial, as its appearance as a dark tissue region depends on the time elapsed after contrast medium (CM) injection.

We hypothesized that using a time series of T1 measurements in these patients can yield tissue characteristics discriminating myocardial necrosis without MVO from necrosis with MVO, thus, providing more quantitative insights into tissue coupling to circulation and allowing for more robust quantification of MVO mass.

Methods:

T1 maps were acquired in 15 patients 3-7 days after STEMI at 1, 3, 6, 9 and 20min after CM injection (0.2mmol/kg Gadovist®, gadobutrol, Bayer Healthcare). At 20min post-injection, late-gadolinium-enhancement (LGE) images were also acquired. The time series of T1 maps were spatially matched for each patient using a non-linear transform (3D Slicer BSpline transform with displacement up to 2mm). The matching was estimated as the transform that brings the left ventricle (LV) on the respective map into a pixel-to-pixel correspondence with the LV on the pre-CM T1 map. Thereafter, the LV myocardium outlined manually on the pre-CM map was subdivided into 3 clusters with k-means algorithm based on the time-series’ intensity. To do so, each pixel of the myocardium was represented as a vector of the values of that pixel in the 5 maps. K-means was initialized with each map’s 40th,60th and 80th histogram percentile. To further consider CM elimination by the kidneys during the 20min, the blood T1 evolution in the LV cavity was also assessed.

Results:

The subdivision followed the T1 differences presented in the post-CM T1 maps. When MVO was discernible on LGE, the clustering approach also divided well the myocardium into the infarcted disconnected from blood i.e. circulation (necrosis with MVO), the infarcted connected (necrosis without MVO) and the normal tissue. In these cases, generally, the T1 evolution of the healthy tissue was monotonously increasing and followed the blood T1 kinetics (same slope, different offset; Figure 1), while the infracted tissue presented 2 different types of T1 evolution over time (Figure 1: i. monotonous decrement or an initial decrement followed by an monotonous increment; ii: monotonous increment following the blood kinetics with different offset). When MVO was not present on LGE, the resulting clusters were commonly presenting T1 kinetics in parallel with the blood pool.

Conclusion:

These preliminary results indicate that the proposed clustering approach applied to times series of T1 maps post CM injection allows to discriminate normal from the necrotic myocardium and to identify regions within the necrotic tissue, which exhibits CM kinetics decoupled from myocardial circulation. This approach may thus permit objective quantification of MVO. Further confirming studies are warranted.

BibTex

Module: