1189

Combined T1, T2, and T2* mapping using a multi-inversion multi-echo spin and gradient echo EPI sequence
Mary Kate Manhard1,2, Congyu Liao1,2, Jason Stockmann1,2, Daniel Park1, SoHyun Han1,2, Jonathan R. Polimeni1,2,3, Berkin Bilgic1,2, and Kawin Setsompop1,2,3

1A.A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, 2Department of Radiology, Harvard Medical School, Boston, MA, United States, 3Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA, United States

Synopsis

Rapid EPI acquisitions are limited by geometric distortion and blurring due to long readouts. Here we incorporate several recent advances in reconstruction approaches and hardware technology to mitigate some of these errors, and implement a multi-echo multi-inversion EPI-based sequence that can be used to find quantitative proton density, T1, T2, and T2* maps. In addition, high quality clinical contrasts can be generated from these maps including T2-, T2*-, T1-, and FLAIR-weighted images. Our protocol provides high-quality, whole-brain, multi-contrast maps with minimal distortion in scan times of 1-3 minutes.

Introduction

Echo-Planar-Imaging (EPI) is one of the most rapid image acquisition schemes in MRI. However, the undesirable geometric distortion and blurring intrinsic to EPI have limited its usage in clinical settings. Recent advances in image reconstruction approaches and hardware technology have shown potential in overcoming these issues. In particular, low-rank reconstruction approaches1,2 have enabled multi-shot EPI with low distortion and blurring to be reconstructed robustly in the presence of shot-to-shot phase variations. Moreover, combined RF and B0 shim array hardware3,4 have demonstrated the ability to significantly reduce B0 inhomogeneity and hence distortion in EPI. In this work, we integrate these new reconstruction and hardware innovations to create high-quality EPI data and develop an efficient EPI-based sequence that can rapidly acquire multi-inversion and multi-echo images. We demonstrate that data from this rapid acquisition can be used to find quantitative maps including proton density (PD), T1, T2, and T2* maps from a single scan. In addition, high-quality clinical contrasts including proton-density, T2-, T2*-, and T1-weighted images can be readily generated from the data series.

Methods

The proposed sequence uses a multi-echo gradient and spin echo (GESE)5,6 acquisition with five echoes per excitation, achieving two gradient echoes, two mixed gradient-and spin echoes, and one spin echo. The multi-echo acquisition is combined with a fast, multiple inversion recovery sequence7, as demonstrated in Fig.1. A non-selective adiabatic inversion pulse is applied before acquiring all slices, reordering the slices every repetition by permuting the slice order by a shift factor (here we used 5). The sequence was tested on two healthy subjects (Siemens Prisma 3T) after written informed consent. The data are acquired with slice-by-slice dynamic shimming using a 32-channel combined RF and B0 shim array (AC/DC coil)4 to decrease the distortion in the phase encoding direction (Fig.2a). Sequence parameters include: FOV=220×220mm, slice thickness = 4mm, 40 slices, first inversion=31ms. Two protocols were implemented:

1) R=4, single shot, 1.2×1.2mm resolution, PF=6/8, TEs=[14 43 87 116 145]ms, TR=8.2s, 8 TIs (65s of acquisition + 32s reference scans)

2) R=8, 3 shots, 1×1mm resolution, PF=7/8, TEs=[16 44 84 112 140]ms, TR=7.5s, 8 TIs (3min of acquisition + 60s of reference scans)

The multi-shot acquisition was reconstructed using MUSSELS1, which has shown great improvement over SENSE (sliding-window) reconstruction (see Fig.2b). The resulting 40 images per slice from each acquisition were fit jointly voxel-wise using Bloch simulations and dictionary matching to achieve PD, T1, T2, and T2* maps. From these maps, clinical contrast images were generated using typical imaging sequence parameters (TI, TE, TR) for each contrast.

Results

Fig.2 demonstrates the advantages of using dynamic shimming and MUSSELS reconstruction to mitigate EPI errors. The dynamic shimming reduces the distortion due to B0 by approximately half, resulting in an 8-fold improvement in distortion using a single-shot R=4 acquisition (Fig2a). MUSSELS reconstruction allows for multi-shot acquisitions with reduced distortion while accounting for phase mismatches between shots (Fig2b). Fig.3 shows representative images from one slice in the single shot R=4 acquisition, with varying inversion times and echo times, that were matched to a dictionary of values to create quantitative paramter maps. Fig.4 shows PD, T1, T2, and T2* maps from representative slices in the R=4 single shot acquisition as well as four clinical contrasts generated from these maps. Fig.5 shows similar maps from the R=8, 3 shot acquisition with dynamic shimming (~16-fold effective distortion reduction) reconstructed using MUSSELS. Some remaining artifacts can be seen in this acquisition, most likely due to motion between shots. Improvements to the acquisition and reconstruction model should help provide cleaner images at this high acceleration factor with negligible distortion.

Discussion

Results show that it is possible to achieve high quality maps with minimal distortion in a short scan time. The sequences demonstrate a potential fast screening protocol with whole brain coverage that include quantitative PD, T1, T2 and T2* maps, in addition to the standard clinical images that are generated from the maps. This acquisition can be improved by optimizing the inversion time scheme8, echo time scheme, and number of echoes and inversions. Future work will look towards faster reference scans based on gradient echo images to further reduce the total scan time. In addition, simultaneous multi-slice (SMS) acquisition9 would allow for more flexibility such as reducing the scan time by 2-4× or achieving up to 1mm isotropic resolution in the same time, while still achieving whole-brain coverage.

Acknowledgements

This work was supported in part by NIH grants: F32EB026304, R01MH116173, R01EB020613, R01EB019437, U01EB025162, P41EB015896, and the shared instrumentation grants: S10RR023401, S10RR019307, S10RR019254, S10RR023043

References

1. Mani M, Jacob M, Kelley D, Magnotta V. Multi-shot sensitivity-encoded diffusion data recovery using structured low-rank matrix completion (MUSSELS). Magn Reson Med. 2017;78(2):494-507. doi:10.1002/mrm.26382.

2. Haldar JP. Low-Rank Modeling of Local k-Space Neighborhoods (LORAKS) for Constrained MRI. IEEE Trans Med Imaging. 2014;33(3):668-681. doi:10.1109/TMI.2013.2293974.

3. Hoffmann J, Shajan G, Scheffler K, Pohmann R. Numerical and experimental evaluation of RF shimming in the human brain at 9.4 T using a dual-row transmit array. Magn Reson Mater Physics, Biol Med. 2014;27(5):373-386. doi:10.1007/s10334-013-0419-y.

4. Stockmann JP, Witzel T, Keil B, et al. A 32-channel combined RF and B 0 shim array for 3T brain imaging. Magn Reson Med. 2016;75(1):441-451. doi:10.1002/mrm.25587.

5. Schmiedeskamp H, Straka M, Newbould RD, et al. Combined spin- and gradient-echo perfusion-weighted imaging. Magn Reson Med. 2012;68(1):30-40. doi:10.1002/mrm.23195.

6. Eichner C, Jafari-Khouzani K, Cauley S, et al. Slice accelerated gradient-echo spin-echo dynamic susceptibility contrast imaging with blipped CAIPI for increased slice coverage. Magn Reson Med. 2014;72(3):770-778. doi:10.1002/mrm.24960.

7. Renvall V, Witzel T, Wald LL, Polimeni JR. Automatic cortical surface reconstruction of high-resolution T1echo planar imaging data. Neuroimage. 2016;134:338-354. doi:10.1016/j.neuroimage.2016.04.004.

8. Cohen O, Polimeni JR. Optimized inversion-time schedules for quantitative T1 measurements based on high-resolution multi-inversion EPI. Magn Reson Med. 2017;00(August):1-12. doi:10.1002/mrm.26889.

9. Setsompop K, Gagoski BA, Polimeni JR, Witzel T, Wedeen VJ, Wald LL. Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planar imaging with reduced g-factor penalty. Magn Reson Med. 2012;67(5):1210-1224. doi:10.1002/mrm.23097.

Figures

Sequence diagram. A non-selective inversion pulse is played once for all slices, and slices are reordered in each repetition, shifted by a factor of 5. During each slice acquisition, five echoes are acquired, 2 gradient echoes, 2 asymmetric spin echoes, and one spin echo.

Clear improvements can be seen in the slice-specific shimmed B0 field map (bottom left) compared to the global second order shim (top left), where the global standard deviation improved from 25Hz to 9Hz (more than 2-fold). Images acquired in both A-P and P-A direction with R=1 are displayed in one slice (middle) so that the distortion can be clearly seen. The R=4 acquisition shows ~8-fold improvement in distortion (bottom right). The AC/DC coil is pictured (top right). B) MUSSELS greatly improves image reconstruction compared to standard SENSE when using multi-shot acquisitions, allowing for a higher acceleration factor with less distortion and shorter echo times.

A) Select images in one slice from different repetitions (which results in different inversion times because of the slice order permutation), with colors corresponding to the repetition time shown. B) Select images at different echo times from the last inversion time shown. In this acquisition, 8TRs × 5TEs resulted in 40 images with different contrasts to be used in the fitting.

A) PD, T1, T2, and T2* maps from three representative slices using a single shot, R=4 acquisition with dynamic shimming applied, resulting in a ~8-fold reduction in distortion. B) From these maps, clinical contrasts were generated including T1-weighted, T2-weighted, T2*-weighted, and T2-FLAIR.

PD, T1, T2, and T2* maps from one representative slice using a three-shot, R=8 acquisition with MUSSELS reconstruction, resulting in an ~16-fold reduction in distortion. Some remaining artifacts due to motion between shots can be seen in this data.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
1189