Fig 1.
Illustration of the reconstruction process consisting in projecting several cardiac cycles with beat-to-beat variations into a mean cardiac cycle.
Left side of the Figure represent three example cardiac cycles (left top is ECG and left bottom is TDI displacement). Below TDI curve are letters corresponding to cardiac phases (C = isovolumic Contraction, S = Systolic ejection, R = isovolumic Relaxation, E = Early diastole filling, D = Diastasis, and A = Atrial contraction). Right side has a top part (2 phase model) showing how the peak E wave positions do not temporally align, whereas bottom part (6 phase model) shows the alignment much better. The two circles represent zoomed part of the TDI displacement curves.
Table 1.
Patients Characteristics (used to collect data for the simulation part).
Fig 2.
Example of myocardial velocity curves in the longitudinal axis and definitions of the six cardiac phases (C, S, R, E, D, and A) used in this study.
C is considered as the beginning of the systole and of the cycle. A is considered as the last part of the diastole and of the cycle.
Fig 3.
Tissue phase mapping MRI simulation with acquisition spanned over different cycles (patient 7).
A template typical velocity curve (A) was used for the simulation. An acquisition spanned over cycles of the same duration results in the velocity curve (B). (C) represents the velocity curves obtained with varying filling order of the k-space. The dark line is the median curve and the +/-2SD space is represented in grey. (D) represents the velocity errors induced by the HR variability.
Table 2.
Characteristics of the MRI simulator.
Fig 4.
Time misalignments induced by Feinstein’s model (Left column) and their decomposition into what is caused by the use of a biphasic model (middle column) and by Weissler’s prediction formula (right column).
(A) The top line plots represent the time misalignments for each position within the 306 cardiac cycle of the database (one black line per cardiac cycle). (B) The middle line plots represent the limit of agreement between each model and the ideal projection. (C) The bottom bar-plots represent the mean limit of agreement within each of the six phases.
Table 3.
Maximum temporal displacements and velocity errors during Feinstein projection.
Fig 5.
Influence of time displacements during Feinstein projection (y-axis) on velocity errors during phase-contrast acquisition (x-axis).
Each point represents a distinct patient. The correlation is good (r2 = 0.7, p = 0.02) despite the small size of the sample.
Fig 6.
Tissue phase mapping velocity curves obtained from healthy volunteers with (A) low RR variations and (B) high RR variations.
The graphs on the top line represent the cardiac cycle duration (Y-axis is the RR interval) during all the acquisition (X-axis is the position of the cardiac cycle during the acquisition). The velocity curves are on the bottom line and the star symbols (*) indicate the isovolumic relaxation peaks. These peaks are more precise (higher peaks) in the group with lower RR variability. SDRR is the standard deviation of cardiac cycle durations during the acquisition.