Background

Myocardial T1 mapping is emerging as powerful tool for tissue characterization, however the presence of intramyocardial or epicardial fat can contaminate T1 values through partial voluming, or preclude analysis, particularly in areas of infarct or thin walled myocardium, such as the right ventricle. We propose and evaluate a new combined fat-water separated saturation-recovery imaging sequence (IDEAL-T1) for water-separated T1 mapping.

Methods

The IDEAL-T1 approach combines a gated, segmented multi-echo gradient recalled echo readout for fat-water separation, based on the "iterative decomposition of water and fat with echo asymmetry and least squares estimation" (IDEAL) method[1], with saturation recovery T1 mapping[2–4]. Images at 4 saturation recovery (TS) times were acquired at a basal slice in diastole over 2 breathholds; one for a non-saturation prepared image, with >4 seconds of recovery between segments, and another for 3 images with incremental TS times. Typical parameters: (Siemens Sonata, 1.5T) TE 2.06, 4.43, 6.8 ms, TR 8.59, flip angle 20°, TS 302-701 ms, FOV 360 × 259 mm, acquisition matrix 256 × 129, phase resolution 70%, 6/8 partial Fourier, 27 views per segments (4 shots per image). Data from water-separated images was scaled by the non-saturated image and fit to a 1-parameter mono-exponential curve, using a Bloch equations simulation look-up table approach to correct for readout-effects on apparent saturation efficiency. In phantom experiments, with a physiologic range of T1 and T2 values (14 phantoms), IDEAL-T1 was validated against an inversion spin-echo sequence. In-vivo evaluation of myocardial T1 was completed in 6 healthy individuals and compared to a single-shot saturation recovery sequence (SASHA)[2] in the left ventricle.

Results

Simulations reveal negligible dependence on T1, T2, and off-resonance (up to 250 Hz), but dependence on B1 errors and saturation efficiency. Phantom experiments show excellent correlation with spin-echo values (R2 0.9996, p < 0.0001) with a mean underestimation of 2.4 ms (Figure 1) and a standard deviation of the difference of 7.4 ms. In vivo evaluation shows a larger underestimation, with a mean difference of -32.5 ms (Figure 1) and a standard deviation of the difference of 12.3 ms. Sample fat and water separated images are shown in Figure 2, where a thin rim of RV fat is revealed on the fat image, and a fat and water profile through the wall illustrates the large region of fat and water overlap.

Figure 1
figure 1

Bland-Altman analysis for phantom (top) and in vivo (bottom) experiments.

Full size image
Figure 2
figure 2

IDEAL-T 1 images showing typical water separated image (top), fat separated image (middle), and signal profile (bottom) across the right ventricle free wall. A small rim of fat is not otherwise visible unless noted on fat separated image.

Full size image

Conclusions

IDEAL-T1 provides the benefit of fat-water separation with quantitative myocardial T1-mapping with a small underestimation in T1. Areas of thin myocardium, including the right ventricle, may benefit from resolving zones of partial voluming with fat by having water only images for analysis.

Funding

The authors acknowledge financial support from CIHR, AIHS, WCHRI.