The LV blood pool signal intensity at the time of peak contrast enhancement was visibly reduced at longer values of TE (). Short-axis and four-chamber long-axis views are shown for the same subject at times of peak RV blood enhancement, peak LV blood enhancement, and peak myocardial enhancement. The time-intensity curves of the LV blood pool ROI show that
effects are only significant at peak Gd concentrations during the first pass (). The undistorted time-intensity curve (, dotted line) estimated for TE = 0 is shown for comparison. The time-intensity curves () for this subject were representative. Measurement of
of the LV blood pool at peak concentration for the full dose was 9.1 ± 3.1 ms (mean ± SD, N
= 10) for the 8-mm-thick short-axis slices.
FIG. 1 Multiecho images for short-axis (left) and four-chamber (right) views at the time of peak RV enhancement (top row), peak LV enhancement (middle row), and peak myocardial enhancement (bottom row) for TE = 1.3, 2.2, 3.1, and 4.1 ms (left to right). Note (more ...)
Time-intensity curves for the LV blood pool ROI for each TE. The TE = 0 curve (dotted line) is estimated based on a least-squares fit to the multiecho data set. The initial two time frames are proton density reference images.
For the seven studies for which there were six measurements (3 orientations × 2 slice thickness values) the
estimate was 9.3 ± 3.7 (mean ± SD) ranging from 5.6 to 21 ms, estimated from all the measurements. A one-way analysis of variance (ANOVA) indicated that the mean study-to-study variation had statistical significance (P
), which contributed to the relatively large variance (see Discussion
). A nonparametric test (Kruskal-Wallis) revealed the same statistical significance. The standard deviation (SD) of the six measurements per study ranged from 0.4 to 3 ms, with a mean value of 1.1 ms. The study with the highest measured SD (3 ms) corresponded to the study with highest
estimate (mean = 17 ms), while the study with the lowest SD (0.4 ms) was one of the studies with a low mean
value (7.3 ms). This implies that the
loss was independent of slice orientation or thickness, and therefore was not due to intravoxel dephasing.
The measured SNR in the LV blood pool used for the Monte Carlo simulation was in the range of 29–72 (56 ± 14, N
= 7) for the 8-mm slice thickness, and 18–37 (30 ± 6, N
= 7) for the 4-mm slice thickness. For an actual value of
= 5 ms, the
estimates for the Monte Carlo simulation had SDs of 0.25, 0.2, 0.11, 0.10, 0.08 ms at SNR = 20, 30, 40, 50, and 60, respectively. For an actual value of
= 20 ms, the
estimates for the Monte Carlo simulation had SDs of 2.5, 1.7, 1.25, 1.0, and 0.8 ms at SNR = 20, 30, 40, 50, and 60, respectively.
Distortion of the AIF was estimated at TE = 0.6 and 1.5 ms relative to the undistorted AIF predicted for TE = 0. For
= 6 ms, signal loss at the peak LV blood pool enhancement ranged between 10% (TE = 0.6 ms) and 23% (TE = 1.5 ms). For
= 12 ms, signal loss at the peak LV blood pool enhancement ranged between 5% (TE = 0.6 ms) and 12% (TE = 1.5 ms). At
= 6 ms, the increase in FWHM was 5% (TE = 0.6 ms) and 12% (TE = 1.5 ms). At
= 12 ms, the increase in FWHM was 2% (TE = 0.6 ms) and 6% (TE = 1.5 ms).