3244

Accelerated multislice MRI with patterned excitation
Jacco A de Zwart1, Peter van Gelderen1, Yicun Wang1, and Jeff H Duyn1
1Advanced MRI section, LFMI/NINDS, National Institutes of Health, Bethesda, MD, United States

Synopsis

Keywords: Pulse Sequence Design, Pulse Sequence Design, accelerated imaging; displacement imaging; diffusion imaging

Motivation: Accelerate spin echo multi-slice MRI.

Goal(s): Combine echo dephasing and rephasing segments of different slices to reduce scan time.

Approach: Use a composite RF pulse to excite upcoming slices while refocusing signal from a current slice.

Results: Implementations at 3 T for spin echo based displacement encoding and spin echo and stimulated echo based DTI show 2-3 fold acceleration.

Impact: The proposed technique can accelerate 2D multi-slice imaging techniques that rely on multiple RF pulses per slice for signal generation and acquisition. It can be combined with simultaneous multi-slice and SENSE approaches for additional acceleration.

Introduction

MRI can generate various contrasts to accentuate aspects of tissue composition, physiology, and structure. A series of RF and magnetic field gradient pulses is played out in sequence, with specific timings, to manipulate contrast and encode spatial information for image generation, often sequentially. Examples are inversion recovery or spin-echo (SE) contrast preparation, which are followed by a signal encoding/collection phase. This allows excellent contrast control but can be time-inefficient. 'Patterned Multislice Excitation' (PME) MRI[1] is a new class of imaging methods that combines contrast generation and signal collection phases to increase efficiency. Application to the measurement of water diffusion[2] and tissue pulsations[3] is demonstrated in human brain at 3 T.

Theory

PME accelerates 2D imaging techniques that require more than one RF pulse per slice by changing each RF pulse's excitation pattern to not only act on the current slice, but to also manipulate magnetization of upcoming slices. The slice pattern targeted with each RF pulse, together with slice acquisition order, dictates the sequence and timing of RF pulses that each slice experiences. Each RF pulse is followed by signal acquisition, increasing time-efficiency.
Figure 1 compares PME-SE with conventional-SE MRI. The latter requires two RF pulse intervals to acquire data for one slice: an excitation pulse and subsequent refocusing pulse, followed by signal acquisition (Figure 1a). With PME-SE, each RF pulse simultaneously excites one slice and refocuses a previously excited one (Figure 1b). A gradient crusher following the RF pulse serves to dephase signal originating from outside the acquisition slice, while refocusing SE signal excited in a previous interval. Pulse shapes, flip angles or slice thicknesses can vary for pulse components. Pulse pattern and crusher gradients can be chosen to extend TE over multiple slice TR (sTR). More than two slices can be excited simultaneously, e.g. to achieve stimulated echo (STE) PME (Figure 2). PME combines a slice-select gradient with a frequency-modulated pulse similar to simultaneous multislice (SMS) MRI[4]. PME and SMS can be combined.

Methods

PME MRI was evaluated on a 3 T Prisma (Siemens Healthineers, Erlangen, Germany) equipped with 32-channel head coil and AS82 gradient set (maximum amplitude: 80 mT⋅m−1; slew rate: 200 T⋅m−1⋅s−1). A pulse-oximeter (MP150, Biopac, Goleta, CA, USA) provided cardiac phase for brain pulsation analysis. Scans were performed on healthy volunteers (n=19, 14 female, aged 29.4±8.8 years) under IRB-approved protocol.
PME-SE and PME-STE implementations used an EPI readout after the crusher, which also served to generate tissue displacement or diffusion contrast. Interleaved slice acquisition with alternating slice select gradient polarity for lipid suppression was used. At slice stack edge, PME pulses were adjusted to cyclically prepare slices at the other end for subsequent volume acquisitions. SE-PME RF pulses were generated by simple addition of 90° excitation and 180° refocusing sub-pulses.
Cardiac-induced vascular pulsations cause subtle brain-tissue movement, which can be measured with SE-based phase-contrast velocity-encoded MRI[5], which PME can accelerate. PME-SE was used with: 150 ms⋅mT⋅m−1 crusher gradients; 4 ms duration RF pulse; single-shot EPI; 2-mm isotropic resolution; rate-3 SENSE; 49 axial-oblique slices (along AC-PC); 25% slice gap; 63-ms TE; 40-ms sTR; 1960-ms volume TR (vTR); 200 repetitions; separate experiments for 3 velocity encoding directions.
Using rate-2 SENSE and otherwise similar parameters[1], PME-SE and PME-STE (TM=5 sTR, Figure 2) were compared to conventional-SE for diffusion imaging with twelve (2.0 and 2.5 mm isotropic resolution) or six (3 mm isotropic resolution) diffusion directions, and two b-values (b~1100 and b<25 s⋅mm−2). For PME-STE, crusher orientation was modulated over sTR to eliminate SE signal. For PME-SE, crusher varied over vTR. TE for PME-SE/PME-STE/conventional-SE was 66.5/50.4/59.8 (3-mm) and 84.0/68.3/70.2 (2-mm) ms; vTR was 1800/1376/3800 (3-mm) and 2444/2068/4900 (2-mm) ms. See Table 1 in [1] for additional details.

Results

Sample results for tissue displacement and water diffusion imaging are shown in Figures 3 and 4&5, respectively. vTR was reduced by 53/52/50% (PME-SE) and 64/61/58% (PME-STE) compared to conventional-SE at 3.0/2.5/2.0-mm resolution. Compared to conventional-SE, relative temporal SNR per unit time at 3.0/2.5/2.0-mm resolution was, for PME-SE: 1.25±0.02/1.16±0.05/1.13±0.03 (@low-b) and 1.38±0.08/1.26±0.08/1.18±0.02 (@high-b); for PME-STE: 0.79±0.03/0.86±0.02/0.82±0.01 (@low-b) and 0.81±0.04/1.00±0.06/0.95±0.02 (@high-b). PME led to increased SAR (due to scan time reduction) and RF pulse amplitude.

Discussion & Conclusion

PME allowed 2-3 fold accelerated SE and STE multi-slice MRI. As with SMS, PME acceleration increases peak RF amplitude and power deposition approximately linearly with acceleration factor. At field strengths of 3 T and above, this may limit the ultimately achievable acceleration.

Acknowledgements

This work was supported by the intramural research program of the National Institute of Neurological Disorders and Stroke, National Institutes of Health.

References

[1]de Zwart JA, van Gelderen P, Wang Y, Duyn J. Accelerated multislice MRI with patterned excitation. Magn Reson Med [in press], https://onlinelibrary.wiley.com/doi/10.1002/mrm.29850

[2]Le Bihan D, Molecular diffusion nuclear magnetic resonance imaging. Magn Reson Q 1991, 7:1-30

[3]Aletras AH, Ding S, Balaban RS, Wen H. DENSE: displacement encoding with stimulated echoes in cardiac functional MRI. J Magn Reson 1999, 137:247-252

[4]Larkman DJ, Hajnal JV, Herlihy AH, Coutts GA, Young IR, Ehnholm G. Use of multicoil arrays for separation of signal from multiple slices simultaneously ecxited. J Magn Reson Imaging 2001, 13:313-317

[5]Greitz D, Wirestam R, Franck A, Nordell B, Thomsen C, Stahlberg F. Pulsatile brain movement and associated hydrodynamics studied by magnetic resonance phase imaging. The Monro-Kellie doctrine revisited. Neuroradiology 1992, 34:370-380

Figures

Figure 1: PME exemplified with two-slice excitation for SE. (a) In conventional SE MRI, two RF pulse intervals ('TR na' & 'TR nb', following 'RF na' & 'RF nb') are used to acquire data for slice n ('Acq n'). Crushers ('Cr na' and Cr 'nb') are used to select SE signal 'SE [na,nb]'. (b) In PME, slice n is refocused while slice n+1 is excited, to be acquired in the subsequent pulse interval. Crusher 'Cr n' suppresses gradient echo signals and leads to generation of a SE signal from pulses in intervals n−1 and n, referred to as 'SE [n−1,n]'. The pattern is shifted by one slice on subsequent repetitions.

Figure 2: Example of stimulated echo (STE) generation with a three-slice excitation pattern 'x0xx'. In addition to the slice under study, an adjacent slice as well as three slice positions over, is excited. As a result, in each pulse sequence interval n, a stimulated echo is generated that has seen pulses at n, n−1, and n−3, indicated as 'STE [n−3,n−1,n]'. The single slice gap ('null excitation') leads to lengthening of the STE mixing time by one slice TR, increasing it from one to two slice TRs.

Figure 3: Tissue displacement per sTR as function of latency with respect to cardiac event for three voxels from a single slice. Anatomical reference and displacement maps for the 0.0-0.1 s latency bin from experiments with gradient in anterior-posterior (AP), left-right (LR), and head-foot (HF) directions are shown (displacement grayscale range was ±25, ±25, and ±40 μm per 40-ms sTR, respectively, for AP, LR, and HF maps). Brain outline in red. For 3 voxels (colored crosses), displacement as function of latency is plotted for a gradient in the AP, LR, and HF directions, respectively.

Figure 4: Sample results for 2-mm resolution patterned multislice excitation (PME) spin-echo (SE) diffusion-weighted (DW) MRI. Top row shows MPRAGE localizer with slice number superimposed. Second row shows the diffusion direction color maps. Third row shows the fractional anisotropy (FA) maps. Bottom row shows the mean diffusivity (MD).

Figure 5: PME DTI MRI with SE and STE (5 sTR mixing time) acquisition schemes compared with conventional-SE DTI MRI. Top left, anatomical localizer; top row, diffusion direction color maps at 3.0-mm isotropic resolution; middle and bottom rows, 2.5-mm and 2.0-mm isotropic resolution. Acquisition time (mm:ss) for the data used is printed below each image. At 3.0-mm resolution, PME-SE MRI allows orientation map acquisition in ~13 s. Equivalent SE MRI requires ~27 s. PME-STE MRI can be performed in ~10 s, but with an approximately 2-fold SNR loss associated with STE generation.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)
3244
DOI: https://doi.org/10.58530/2024/3244