Several studies have shown the correlation between left ventricle (LV) rotation and cardiovascular diseases^1-3. Regional parameters such as regional rotation are studied as markers of cardiac disorders⁴. So far, a large number of methods have been proposed to investigate LV rotation. Among the cardiovascular MR imaging sequences, Tagged MRI (tMRI) is a promising technique for this purpose⁵. Polar tMRI (Radial and Circular tagging), due to its compatibility with LV anatomy is the method of choice for rotation and strain quantification since it adapts best to the cardiac geometry and motion⁶. Several algorithms were developed to estimate the cardiac motion from tMRI among them phase-based methods are well known because of robustness against fading effect in tMRI⁸. Monogenic signal method is a phase-based and robust procedure that is not performed on radial pattern tMRI thus far.

Monogenic signal is capable of estimating local parameters of a gray-valued image such as local phase, local amplitude and local orientation by a set of appropriate quadrature filters⁷. The proposed rotation calculation is divided in two fundamental steps: First, estimation of displacement vector between two frames using Monogenic signal approach⁸; Second, calculation of rotational motion around the LV center by averaging of tangential component of displacement vector at desired myocardium region. Figure 1 shows the stages of displacement vector estimation. For every frame, the Monogenic signal parameters was calculated. Next, the optical flow algorithm for two frames, with monogenic phase-constancy assumption instead of the well-known brightness-constancy, was performed⁸. Also, for evaluation of proposed method, a phantom tMRI frame, at several known rotation (-5 to 7 degrees) was simulated and Mean Square Error (MSE) was calculated. In this study, we recruited 16 healthy volunteers and for each subject radial tagged MRI at Mid-level was applied. Regional rotation is depicted on a 6-segment standard division (referred to 6 colored regions of figure 1) according to American Heart Association (AHA). Further, the global and transmural changes of rotation were calculated.

Figure 2 demonstrates rotation plots for 3 transmular layers of LV , namely Sub-Endocardium, Mid and Sub-Epicardium layers, as well as the global rotation throughout cardiac cycle for a typical healthy subject (Left). It also shows the whiskers plot (Box plot) of rotation peaks for the aforementioned areas (Right). As can be seen in figure 2, the peak rotation decreased across the wall from Endo to Epi at LV mid-level. This fact is concordant with previous studies⁹. Figure 3, also illustrates the same whiskers plot parameters, but at the 6 regions in the circumferential direction. The slope of peak rotation mean in transmural changes direction is -0.33 degree per millimeter. Also, the calculated MSE for phantom frame was obtained less than 0.02.The ability of radial tMRI in revealing the rotational motion in combination with proposed method seems a promising approach to reliable estimation of rotational related factors. The robust and reproducible pixel-wise displacement vector estimation by proposed method, provides an opportunity to accurately analyze the LV twist and its effect on the cardiac function. Application of the method on 16 healthy volunteers showed that the peak rotation has some variations across the 6 standard segments of the LV, nevertheless it showed a very robust pattern for transmural changes of the rotation across the wall, calculated as -0.033 degree per millimeter. This factor is then hypothesized to act as a new marker besides other parameters for cardiac evaluation.