Fig 1.
(A) The TD scout was acquired during free-breathing with cardiac synchronization. Images were acquired every two heartbeats to allow for signal recovery (TR≈2s). (B) The cDTI acquisitions was cardiac and respiratory triggered. One image was acquired every respiratory cycle (TR≈4s). (C) Details of the cDTI sequence used in this work. The timings of the EPI acquisition Tnav/δ90°/δ180°/Ta/Tb/TEPI-echo/TEPI/TE were 8/2/4/24/20/9/23/61ms and the timings of the repeated diffusion encoding gradient δ1/δ2 were 7.2/12.6ms. The deadtime δd was 4ms.
Fig 2.
Calculation of HA, TA, and E2A at each cDTI voxel.
(A) Stack of Nb diffusion weighted images with an example LV ROI. The epicardial surface is shown in blue and the endocardial surface in red. (B-C) RGB colormap of the primary and secondary eigenvector orientation from the reconstructed diffusion tensors in the mid LV SA slice. (D-F-H) HA, TA, and E2A were calculated at each voxel after projection of the primary and secondary eigenvectors (E1 and E2) [2, 3, 22]. (E-G-I) HA, TA, and E2A maps in a representative volunteer (E2A is reported as absolute value).
Fig 3.
Examples of TD scout acquisition and corresponding signal as a function of TD normalized over the RR interval.
Representative cases in which it was possible (A) and not possible (B) to acquire cDTI during diastasis. The signal is reported as the median DWI signal (solid line) and interquartile range [Q1 Q3] (dashed line). The signal is computed across all images per volunteer after segmentation (see S1 Fig) and after normalization by the cardiac phase with the mean highest signal. The colored boxes represent the temporal windows in which the cDTI has been acquired. Thus, the width of each box corresponds to the time from triggering to TE normalized with respect to the RR interval. Panel (C) shows the overall signal measured across all volunteers as a function of TD normalized over the RR interval. Median (solid line) and [Q1 Q3] range (dashed line) TD scout signals measured across all volunteers as well as individual values (points) corresponding to each volunteer. Boxplots show the distribution of TDs for early systole, late systole, and diastasis across all volunteers.
Fig 4.
Reconstructed diffusion tensor data at early systole, late systole, and diastasis for a representative volunteer.
From top to bottom: RGB map of the primary eigenvector (E1) orientation, HA map, TA map, and E2A map (E2A is reported as absolute value).
Fig 5.
Transmural HA distributions across six mid-ventricular AHA segments in early systole, late systole, and diastasis in a representative volunteer.
Bottom row shows combined HA distribution across all AHA segments. Rightmost column shows the median profile of the HA distribution colored by cardiac phase. Crosses and hash marks denote the median and interquartile range of HA at a single transmural depth. Note the different HA distributions across the AHA segments, and the complex HA distribution due to the RV insertion points (red boxes) in the anterior and inferior segments.
Fig 6.
3D representation of myofiber orientations at early systole, late systole, and diastasis for a representative volunteer.
(Top) Primary eigenvectors colored by HA. (Bottom) Zoomed-in view of the septal wall. Note the steeper endocardial HA at late systole.
Table 1.
Helix angle and helix angle range per cardiac segments.
Table 2.
Delta helix angle range per cardiac segments.
Fig 7.
Angle statistics at early systole, late systole, and diastasis across all volunteers.
Each volunteer corresponds to a circular marker. Median and IQR are reported across volunteers. (A) Helix Angle Range (HAR), (B) Transverse Angle (TA), and (C) E2 Angle (E2A) [** p-value < 0.01].
Fig 8.
Tensor invariants statistics at early systole, late systole, and diastasis across all volunteers.
Each volunteer corresponds to a circular marker. Median and IQR are reported across volunteers. (A) Mean diffusivity (MD), (B) Fraction of Anisotropy (FA), and (C) ratio of second and first eigen values (λ2/1) [** p-value < 0.01].