Rui Guo1, Zhensen Chen1, Jianwen Luo1, and Haiyan Ding1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, People's Republic of
Synopsis
In this study, we developed a free-breathing,
three-dimensional (3D) T1 mapping method at 3T based on the saturation-prepared,
inversion-prepared radiofrequency-spoiled gradient echo (SPGR) sequence, which
achieved 3D T1 maps with whole heart coverage. Excellent correlation for the T1
values typically observed in myocardium pre and post contrast was obtained
with little residual (R2 = 0.99) and a regression slope 0.93. Homogeneous T1
maps were obtained from in-vivo study. Whole left ventricle T1 mapping was finished
in around ~10 minutes with about 40% navigator efficiency without using any
parallel imaging. Measured normal myocardium T1 values were comparable to those
previously published.Target audience
Scientists and clinicians who are
interested in myocardial tissue characterization.
Introduction
Myocardial T1 mapping is a non-invasive and important tool
for quantitative myocardium characterization, which has been shown great
potential in comprehensive disease assessment[1]. Nowadays, myocardial T1
mapping methods, such as MOLLI[2], shMOLLI[3], SASHA[4] and SAPPHIRE[5], are
generally implemented with 2D single shot imaging technique along with breath
hold, resulting in limited image coverage and imaging resolution. Besides, due to use of long acquisition
window, the cardiac motion and imperfect breath holding may cause image blurring and misregistration among the T1-weighted images, which finally results in the degraded quality on T1
maps[6].
In this study, we developed a free-breathing,
three-dimensional (3D) T1 mapping method at 3T based on the
saturation-prepared, inversion-prepared radiofrequency-spoiled gradient echo
(SPGR) sequence, which could largely overcome the limitations of conventional 2D techniques.
Methods
All imaging studies were performed on a 3T MR system
(Achieva TX, Philips Healthcare, Best, Netherlands). Cardiac studies used a
32-channel phased array (InVivo, Gainesville, Florida, USA) and phantom studies
used an 8-channel head coil.
Sequence design: A
series of 3D volumes were
acquired in an interleaved fashion during diastolic cardiac phase, as shown in Figure 1A. Such interleaved acquisition enabled reducing misregistration among the volumes. Four different T1 weightings were achieved by using non-selective saturation
prepulses with different delay (TSAT) to SPGR readout. An additional
inversion prepulse with a short delay (TINV) was inserted right before the readout of the first volume imaging, for purpose of extending the magnetization T1 recovery dynamic range. In
order to achieve the fully recovered magnetization, five or six idle heartbeats were skipped before the last
volume was acquired. To remove time limits imposed by breath-holds
and achieve desired spatial resolution and coverage, a pencil beam respiratory navigator
was used for respiratory compensation. As for postprocessing of this proposed sequence, pixel-wised 3D T1 maps were calculated using a 3
parameters model: s(i) = Aα(TINV, T1)+Bβ(TSAT,i, T1)*exp(Td/T1),
where i was the ith T1-weighted volume,Td was TINV or TSAT, i (depending on whether there was an inversion prepusle or not in the corresponding imaging volume).
Phantom study: Gel phantoms (n=13) were prepared using Gd-DTPA
(Magnevist, Berlex) and Agarose (Sigma-Aldrich) to approximate T1 values between
300-2200ms. Reference T1 values were measured using the IR prepared single-echo
spin echo sequence (IR-SE) with TI ranging from 100ms to 3000ms. A TR of
10s was used in order for complete magnetization recovery. Imaging voxel size was 2×2×8mm3 and a simulated heart rate 60bpm was employed. Correlation between the T1 values from the 3D T1 mapping and that from spin
echo was performed.
Normal
Human Subjects Study: Under the local Institutional
Review Board approved protocol, 3 healthy volunteers without
known cardiovascular dysfunction were enrolled and scaned with the proposed 3D T1 mapping sequence.
Typical imaging parameters were: FOV 300×300×80mm3,
voxel size 2×2×8mm3, FA 18°, TR/TE 3.28/1.0ms with a 5/8
fractional readout, acquisition window about 100ms. Total scan time for T1 mapping
was around ~10minutes with about 40% navigator efficiency and without using any
parallel imaging.
Results
Excellent
correlation betwwen the proposed and reference method for the T1 values typically observed in myocardium pre and post contrast imaging was obtained (R
2 = 0.99 and regression slope 0.93, Figure 2). Homogeneous T1 maps were obtained from in-vivo study. Representative short
axis view (SAX) T1 map from one healthy volunteer was shown in Figure 3A. Ten SAX T1 maps
covering the whole left ventricle were reformatted to long axis view as shown in Figure 3B. Mean and standard deviation of T1 values obtained with the proposed sequence in normal myocardium was1508.4±34.5ms.
Conclusion
In
this study, we presented a novel 3D free-breathing T1 mapping sequence at 3T that
achieves whole heart coverage with excellent homogeneity. Measured normal
myocardium T1 values are comparable to those previously published [7,8]. The
3D T1 maps should allow for higher resolution myocardial tissue
characterization. By combining with parallel imaging or the state-of-the-art
fast imaging techniques, higher spatial resolution and more efficient imaging
acquisition could be achieved in the future.
Acknowledgements
We would like to thank Dr. Daniel Herzka from the Department of Biomedical Engineering, Johns Hopkins School of Medicine
for his valuable advices and encouragement during the sequence development. References
1. Salerno M, et al. JACC CVI. 2013.
2. Messroghli
DR, et al. MRM. 2004.
3. Piechnik SK, et al. JCMR 2010.
4. Chow K,et al. MRM. 2013.
5. Weingärtner S, et al. MRM. 2013
6. Kellman P, et al. JCMR. 2014.
7. Stanisz GJ, et al. MRM. 2005.
8. Chow K, et al. MRM. 2015.