Radial cardiac T₂ mapping with alternating T₂ preparation intrinsically introduces motion correction

Background

T₂ mapping through variation of the T₂ preparation (T₂Prep) duration has been increasingly used to robustly detect and quantify cardiac edema (Giri et al., JCardiovMagnReson2009). However, if images with incremental T₂Prep duration are acquired in a sequential fashion, irregular breathing patterns and heart rates may adversely affect the quality of the T₂ maps due to misalignment of the source images. A logical alternative is then to acquire all images in an alternating manner (Figure 1ab), where the T₂Prep duration changes cyclically from one heartbeat to the next. Combined with a radial signal readout, this may minimize the vulnerability to respiratory or RR variability. We therefore simulated, implemented and tested the utility of an alternating magnetization preparation approach to T₂ mapping.

Figure 1

a) Schematic of the alternating acquisition pattern. The T₂Prep duration is alternated between 60, 30 and 0 (no T₂Prep) ms from heartbeat to heartbeat. All images are acquired in an interleaved fashion and on average experience similar motion. b) Schematic of the conventional sequential acquisition pattern. The T₂Prep duration is changed only after acquisition of an image. This approach may be more vulnerable to irregular heart rates or respiration patterns. c) Simulations of stability of fitted T₂ values for both methods against heart rate variation. For an input T₂ of 45 ms, both the alternating acquisition (solid line) and the sequential acquisition (dashed line) result in a T₂ variation of ~3 ms over the range of physiological heart rates.

Methods

A navigator-gated ECG-triggered radial gradient-recalled-echo pulse sequence (20 lines per heartbeat, trigger every 3 heartbeats) was implemented to obtain source images for the T₂ maps (van Heeswijk et al., JACCCardiovImag2012), with the possibility to apply the T₂Prep durations of 60/30/0 ms in both an alternating and sequential manner. Bloch equation simulations were performed in order to estimate the fitting residual due to T₁ relaxation (van Heeswijk et al., JACCCardiovImag2012) as well as the accuracy over a range of heart rates. The sequences were validated at 3T (12-channel surface coil array, on a Magnetom Trio, Siemens, Erlangen, Germany) in agar-NiCl₂ phantoms by comparing the resulting T₂ maps to gold-standard spin-echo (SE) T₂ measurements. A mid-ventricular short-axis T₂ map was then acquired with both pulse sequences in 7 healthy adult volunteers. The myocardial surface area was measured in the T₂ maps, while a Student's t-test was applied to detect differences in T₂ values and surface area.

Results

The alternating sequence was as robust to heart rate variation as its sequential counterpart (Figure 1c), while its accuracy was confirmed in the phantoms (T₂ = 45.4 ± 0.7 ms for the alternating method, vs 45.3 ± 0.7 ms for the sequential method and 45.1 ± 0.7 ms for the gold-standard). The myocardial surface area was increased in the alternated T₂ maps of the volunteers (128 ± 24 cm² vs. 111 ± 20 cm², p = 0.04) (Figure 2), while the average midventricular T₂ value slightly differed between the alternated and sequential methods (T₂ = 37.6 ± 6.6 ms alternated vs. 40.4 ± 6.1 ms sequential, p = 0.01).

Figure 2

T ₂ map of volunteer acquired with the alternating (a) and sequential (b) method. Note that consistent with the quantitative findings, the antero-lateral myocardium is thicker when acquired with the alternating method (arrows).

Conclusions

We successfully implemented and tested a T₂ mapping methodology in which magnetization preparation is alternated. The in vivo T₂ maps demonstrate that the alternated acquisition intrinsically aligns its sources images, resulting in a larger available myocardial surface, which in turn may allow for more accurate T₂ value quantification.

Funding

N/A.