Sector‐wise golden‐angle phase contrast with high temporal resolution for evaluation of left ventricular diastolic dysfunction

Figure 1. Schematic overview of the proposed sector‐wise golden‐angle (SWIG) method. A, The k‐space lines are acquired continuously with an angular increment determined by the golden‐section division of each sector. B, Each sector of the radial k‐space is acquired in a separate heartbeat, meaning that there is no mixing of heartbeats that can lead to clustering after physiological resorting. C, The shared‐velocity encoding, in which every other k‐space line is acquired with alternating velocity‐encoding (VENC) signs enables reconstruction of a unique image of each TR using a sliding‐window technique. Each box indicates an image with either positive or negative VENC. Each line indicates a pair of positive and negative VENC image frames that makes up a phase‐subtracted image frame

Figure 2. A, Representative example from 1 image frame from 1 subject shows the difference in sampling uniformity after physiological resorting, together with corresponding point spread function (PSF). B, Analysis of peak PSF value and PSF energy as a distance of radius shows lower peak PSF and PSF energy for SWIG. The curves correspond to the mean over all reconstructed image frames, for all subjects in the study. The horizontal axis corresponds to the full FOV

Figure 3. A, Illustrative example of images from the SWIG method, displayed as phase images, in 3 myocardial phases corresponding to s′, e′, and a′. B, Time‐velocity curves, measured in a single voxel, at different temporal footprints. C, At the narrowest temporal footprint, corresponding to 4 spokes (27.2 ms), the time‐velocity curve is rather noisy. For the next temporal footprint of 6 spokes (40.8 ms), the curve is smoother but reaches nearly the same peak velocity in s′, e′, and a′. Here, 14 and 16 TRs were also included to further illustrate the loss of peak velocity as the temporal footprint is increased. The enlarged view of the e′ peak (location indicated by the black box) further illustrates the differences in temporal smoothing with increasing temporal footprint

Figure 4. A and B, Pulsed‐wave Doppler measurements acquired using conventional clinical methods. C and D, The corresponding measurements derived from MRI with the proposed pulse sequence. All measurements were performed in the same subject as in Figure 3, a 49‐year‐old male with normal left ventricular systolic function and normal myocardial tissue characterization by cardiovascular MRI (CMR), although with slightly impaired diastolic function as seen by both Doppler and the proposed pulse sequence

Figure 5. Correlation plot (A and C) and Bland‐Altman plot (B and D) of myocardial and mitral peak velocities measured with SWIG and pulsed‐wave Doppler echocardiography. Blue circles indicate peak myocardial tissue velocity in systole (s′), orange triangles indicate peak tissue velocity (sign switched) in early filling (e′), green squares indicate peak tissue velocity (sign switched) in late filling (a′), red stars indicate peak transmitral blood flow velocity during early filling (E), and purple diamonds indicate peak transmitral blood flow velocity during late filling (A). Tissue velocities were measured in both the lateral and septal wall of the left ventricle

Figure 6. Scatter plots showing the Bland‐Altman bias from Figure 5 in relation to heart rate and heart‐rate variability defined as the SD of the RR intervals during myocardial velocity (A and B) and mitral flow (C and D) measurements. Blue circles indicate peak myocardial tissue velocity in systole (s′), orange triangles indicate peak tissue velocity (sign switched) in early filling (e′), green squares indicate peak tissue velocity (sign switched) in late filling (a′), red stars indicate peak transmitral blood flow velocity during early filling (E), and purple diamonds indicate peak transmitral blood flow velocity during late filling (A). The dashed lines indicate linear regressions

Figure 7. Residual phase errors before the phase‐error correction. A spherical phantom, 240 mm in diameter, was scanned with a conventional Cartesian phase contrast pulse sequence (A and C) and with the proposed SWIG sequence (B and D) with VENC = 30 cm/s and 150 cm/s