The feasibility of a novel limited field of view spiral cine DENSE sequence to assess myocardial strain in dilated cardiomyopathy

Fig. 1. Novel accelerated zonal spiral cine DENSE sequence. The first RF pulse (red) reduces the y FOV to the region between the red lines, while the second RF pulse (blue) reduces the perpendicular FOV, leaving a region around the heart to image. The schematic sequence diagram also shows the R wave of the ECG, fat suppression (fat sat.), the DENSE encoding and spiral gradients (green) and the slice-selective RF pulses (purple) with increasing flip angle used for imaging. In our pilot study, a mean of 27 frames were acquired (30 ms each), which equates to 54 spiral readouts in each cardiac cycle (TR = 15 ms)

Fig. 2. Example regions of interest for the assessment of strain. Top line shows short-axis (SAX) image with endo-epicardial contour for assessment of radial and circumferential strain. Lower panels show SAX, vertical long-axis (VLA), and horizontal long-axis (HLA) image with contour lines shown in orange for assessment of contour strain. The left-hand images in each panel are magnitude images and the two right-hand images are phase images, where the displacement of each pixel is encoded in the image phase (x and y directions). The SAX images are cropped to a FOV of 110 × 100 mm2 and the long-axis images are cropped to a FOV of 130 × 130 mm2

Fig. 3. Measuring the in-plane excitation profile. Magnitude DENSE image acquired with a 50 mm stimulated echo FOV and a 250 mm readout FOV (cropped to 125 mm) (a). Excitation profiles, averaged over the red (b) and blue (c) regions shown in a. For the purposes of plotting intensity profiles, intensities were normalised to the intensity at the centre of the FOV. Profiles were measured from left to right (b) and from top to bottom (c). The receive coil spatial sensitivity variation causes the slope across the central region of the profiles. Receive coils were placed above (flexible matrix array) and below (spine coil, built into bed) the phantom (top and bottom of a) and sum of squares coil combination was used

Fig. 4. Cine DENSE analysis. Examples of studies in which cine DENSE data analysis was not possible due to insufficient signal-to-noise ratio (top panels) and off-resonance due to a change in magnetic susceptibility (bottom panels). The orange and blue arrows highlight the poor SNR in the lateral wall, which approaches background noise levels (red arrow). A susceptibility or off-resonance artefact often affects the short-axis data (purple arrow). The lower phase image shows phase wrap as the frames shown are acquired near peak systole. The DENSE processing software used here unwraps the phase

Fig. 5. Summary strain–time curves for DCM and control cohort. The plots show the median longitudinal (horizontal long axis, HLA; vertical long axis, VLA) strain, short-axis (SAX) contour strain, and radial and circumferential strain in healthy volunteers (blue curves) and the DCM cohort (red curves). Curves are shown with the inter-quartile range (dashed lines). Peak strain is reduced in all dimensions in DCM patients compared to controls

Fig. 6. Relationship between LVEF and strain. Relationship between left ventricular ejection fraction (LVEF) and contour strains (top 3 panels) and radial and circumferential strains (bottom two panels). Univariable linear regression lines are plotted with 95% confidence intervals. The slope of the regression and the p value are shown above the plots. Points are colour coded by diagnosis; red = DCM, blue = healthy volunteer. The grey bands indicate 95% confidence intervals for the linear model

Fig. 7. Correlation between strain and CMR parameters. Correlation matrix showing the size of the correlation (from − 1 to + 1) between LVEF, mean wall thickness (WT), left atrial (LA) area, right ventricular ejection fraction (RVEF) and strain, in DCM patients only. Only correlations with a p value < 0.05 are displayed. The larger the circle and the more intense colour indicate a stronger correlation. Overall, these results suggest that radial strain in this cohort is a less robust marker for the assessment of LVEF function compared to circumferential, longitudinal, and short-axis contour strain