Rapid and Robust Method for Quantifying Left Ventricular Twist and Torsion from Tagged Cardiac MRI Studies

Cardiovascular disease remains the number one cause of death in the world(1). Quantitative assessment of cardiac function with magnetic resonance imaging (MRI) has the potential to better inform us about changes to heart function during the development of disease, to guide clinical decision making, and to evaluate the effectiveness of therapy. The measurement of left ventricular (LV) twist and torsion in diseases such as dilated cardiomyopathy(2-7) and aortic stenosis(8,9) is increasingly used as a quantitative measure of LV dysfunction. LV twist is a measure of the rotation of the apex relative to the base of the heart. In patients with dilated cardiomyopathy, for example, twist is reduced(4-7) and, in some cases, apical rotation is reversed(5). Alternately, in aortic stenosis twist is increased(8,9). LV twist may serve as a useful method for quantifying the severity of systolic and diastolic dysfunction, especially in the context of preserved ejection fraction. Noninvasive MRI methods such as myocardial tagging, which labels the myocardial tissue with lines that deform with the heart, are established for quantifying LV twist(2,3,8-11). However, the ‘gold-standard’ method for quantifying LV twist, FindTags(12) is time consuming and commercially available techniques can be costly. We have recently validated a rapid quantitative method termed FAST (Fourier Analysis of STimulated echoes) which utilizes tagged cardiac MRI and exploits the fact that object rotation in image space directly corresponds to a rotation in Fourier space for the quantification of LV twist(14).

The FAST method rapidly quantifies LV twist based on four sets of short-axis tagged cardiac MR images (at the apex and the base, with vertical and horizontal tags)(14). The only user interaction needed for FAST processing is contouring of the LV epicardium at the basal and apical slice level in the first cardiac frame. The contour was used to create a mask, which defined the region of interest (i.e. LV myocardium), and eliminated tissues that did not rotate. Subsequent to masking, the tagged image was 2D Fourier transformed and the center peak (FID) in the Fourier image was nulled with a circular mask to reduce the interference during the detection of object rotation. Next, the Fourier image was cropped so only the central 64 lines of k-space remained which reduced image size and processing time. The Fourier magnitude data for each frame was 2D cross-correlated with a rotated version of the frame immediately after it within the same slice. The maximum of the two-dimensional cross-correlation from all tested rotations defined the angle of rotation between those frames. The cumulative inter-frame rotation for any time point defined the net rotation. This process was repeated for all frames in the basal and apical slices. Finally, twist was calculated as the difference in the angle of rotation for matching frames between the apical and basal slices. Peak twist was defined as the maximum twist value. FAST functionality has been incorporated into an open-source Osirix plug-in. Validation of FAST against FindTags was performed in 8 healthy subjects and 6 canines. The intraobserver coefficient of variation(CV) for peak systolic twist (2.9% and 2.6%) and interobserver CV (4.3% and 4.2%) were all small, determined by two investigators performing contouring twice. Linear regression analysis of the FAST and FindTags twist values indicated very good agreement in healthy subjects (R=0.91) and in canines (R=0.95). Bland-Altman comparison of the FAST and FindTags twist results indicated excellent agreement in healthy subjects (bias of 0.02°,95% confidence intervals (-3.7°,3.7°)) and canines (bias of 0.2°,95% confidence intervals (-2.7°,3.1°)). Peak systolic twist healthy subjects (n=20) averaged 10.5±1.9° degrees, which matches well with literature values.

The demonstration will include an introductory presentation on LV anatomy with emphasis on epicardial and endocardial boundaries in short-axis MR images. LV twist will be defined and an algorithm for routine clinical evaluation of twist based on the current literature will be presented. Next, an overview of FAST image processing will be presented with emphasis on the robustness of the method and potential pit-falls. Thereafter, a brief demonstration of how to use the plug-in will be performed. Two laptops with tagged MRI case examples and Osirix will be available for attendees to test out the FAST plug-in.

1 AHA. 2010 Annual Report. 2 Shaw. Int J Cardiol 2008;130:319-325. 3 Chung. J Am Coll Cardiol 2006;47:384-390. 4 Liu. Echocard;27:400-405. 5 Popescu. Eur J Heart Fail 2009;11:945-951. 6 Mornos. Hellenic J Cardiol;52:299-306. 7 Pacileo. Eur J Echocard;12:841-849. 8 Stuber. Circ 1999;100:361-368. 9 Nagel. Eur Heart J 2000;21:582-589. 10 Li. JCMR 2009;11:40. 11 Delhaas. MRM 2004;51:135-139. 12 Guttman. IEEE Comput Graph Appl 1997;17:30-38. 14* Reyhan. JMRI 2012;35:587-93.

LL-INS-SU2A

Rapid and Robust Method for Quantifying Left Ventricular Twist and Torsion from Tagged Cardiac MRI Studies