Multiband DREAM: Multi-Slice B1+ Mapping in a Single Shot
Kay Nehrke1, Arthur Felipe Nisti Grigoletto Borgonovi2, and Peter Börnert1,3

1Philips Research, Hamburg, Germany, 2Philips Healthcare, Best, Netherlands, 3Radiology, LUMC, Leiden, Netherlands

Synopsis

Simultaneous multi-slice imaging has been employed for the DREAM B1+ mapping approach, allowing a multi-slice B1+ map to be acquired in a fraction of a second. Basic feasibility has been shown in experiments on phantoms and in vivo using a clinical 3T MRI system. The presented approach potentially allows free-breathing multi-slice B1+ mapping, freezing respiratory motion in a short single shot acquisition window.

Introduction

In high field MRI, fast and robust in vivo B1+ mapping is an essential prerequisite for many quantitative or parallel-transmit-based applications (1-3). However, most B1+ mapping techniques are too slow for a seamless integration into the clinical workflow. Recently, the DREAM (Dual Refocusing Echo Acquisition Mode) B1+ mapping approach has been introduced (4), allowing the acquisition of a 2D B1+ map in a fraction of a second. Moreover, a volumetric DREAM B1+ map can be acquired in a couple of seconds by successive acquisition of multiple slices. In the present study, the DREAM approach is further accelerated by employing simultaneous multi-slice imaging (5), allowing a volumetric B1+ map to be acquired in a single shot. The new approach has been implemented on a clinical scanner, and basic feasibility has been shown on phantoms and in vivo.

Theory

For multiband slice excitation, sinusoidally modulated RF pulses are employed, representing a superposition of two or more offset frequencies, each one corresponding to a certain imaging slice (Fig.1). Using parallel imaging, the different slices can be unfolded from the acquired data in reconstruction. Nevertheless, the strongly modulated broad-band RF pulses potentially suffer from long pulse duration due to the limited available peak B1+ and SAR constraints at higher field strength. This turns out to be especially a problem for B1+ mapping, because these sequences are generally based on large flip angles required for B1+ encoding. On the other hand, some B1+ mapping approaches separate the B1+ encoding from the spatial encoding. For instance, the DREAM sequence uses a STEAM preparation sequence for B1+ encoding and a low angle pulse train for spatial encoding (Fig.2a), were both, an FID and a STE, are measured as separate recalled gradient echoes under a single readout gradient lobe. Hence, simultaneous multi-slice DREAM B1+ mapping can be implemented combining a non-selective STEAM preparation with a multiband imaging RF pulse train (Fig.2b).

Methods

Phantom and in vivo experiments on the abdomen were performed on a clinical dual-transmit 3T MRI system (Ingenia, Philips Healthcare, Best, The Netherlands). DREAM Multiband was implemented in the scan and reconstruction software of the scanner and used for simultaneous B1+ mapping of four equidistant slices (transversal orientation, slice thickness: 10 mm, slice gap: 40 mm, FOV = 530x460mm², 2D scan matrix = 64x56, STEAM/imaging flip angle = 50°/10°, TEFID/STE=2.3/3.3 ms, TR = 4.5 ms, receive coil: 28-channel anterior-posterior coil array). To improve the stability of SENSE unfolding (g-factor), neighbouring slices were shifted by a quarter field-of-view, which was achieved by appropriate modulation of the the RF pulses (5). The B1+ maps were acquired in a single shot (duration ≈ 250 ms), which allowed acquisition of the in vivo data during free-breathing.

Results

Figure 3 and 4 show the acquired B1+ maps along with the underlying source images of the DREAM sequence for the water-filled body phantom and for the abdominal acquisitions, respectively. The B1+ maps show the well-known shading pattern typical for quadrature excitation at 3T. The local torso SAR level was slightly increased from 9% (conventional multi-slice acquisition) to 13 % (simulataneous multi-slice), which is still very moderate, especially for a B1+ mapping scan.

Discussion

Multiband DREAM allows for multi-slice B1+ mapping in a fraction of a second. Thus, motion is frozen, and the acquisition can be performed during free breathing capable also to capture dynamic processes (like respiration). The presented approach represents an alternative to 3DREAM (6), which is based on a single shot 3D DREAM acquisition. Multiband DREAM provides the advantage of a reduced partial volume effect, because thinner slices may be chosen. Moreover, oversampling in z-direction is not required, increasing the sampling efficiency of the approach. Finally, the approach is more flexible with respect to optimizing the g-factor for the chosen coil array by selecting appropriate phase shifts between the slices.

Acknowledgements

No acknowledgement found.

References

1. Hoult DI and Phil D. JMRI 2000;12:46-67.

2. Katscher U et al. MRM 2003;49:144-50.

3. Zhu Y. MRM 2004;51:775-84.

4. Nehrke K. and Börnert P., MRM 2012;68:1517-26.

5. Breuer, F. A et al, MRM 2005; 53), 684–91.

6. Brenner et al. Proc ISMRM 2014#1455.

Figures

FIG. 1. Multiband Excitation. Simultaneous excitation of multiple slices is achieved by sinusoidal modulation of the RF waveform (example for two slices).

FIG. 2. DREAM sequence, consisting of non-selective STEAM preparation (left) and slice-selective imaging sequence (a: single-slice, b: simultaneous multi-slice).

FIG. 3. Phantom experiments: Multiband B1+ mapping experiment (DREAM). In (a), the positions of the four slices are indicated by blue lines. In (b-d) the corresponding DREAM are shown (top to bottom: four slices). The B1+ mapping scans show the underlying source images (b: FID, c: STE) and the resulting B1+ map (d). Multiband-separation is obvious.

FIG. 4. In vivo experiments: Multiband B1+ mapping experiment (DREAM). In (a), the positions of the four slices are indicated by blue lines. In (b-d) the corresponding DREAM data are shown (top to bottom: four slices). The B1+ mapping scans show the underlying source images (b: FID, c: STE) and the resulting B1+ map (d).



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4271