0262

Simultaneous B1 and T1 Mapping Using Spiral Variable-Flip-Angle Acquisitions for Whole-Brain Coverage in Less Than One Minute
Rahel Heule1,2,3, Josef Pfeuffer4, Craig H. Meyer5, and Oliver Bieri1,2

1Division of Radiological Physics, Department of Radiology, University of Basel Hospital, Basel, Switzerland, 2Department of Biomedical Engineering, University of Basel, Basel, Switzerland, 3High Field Magnetic Resonance, Max Planck Institute for Biological Cybernetics, Tübingen, Germany, 4Siemens Healthcare GmbH, Application Development, Erlangen, Germany, 5Department of Biomedical Engineering, University of Virginia, Charlottesville, VA, United States

Synopsis

Rapid variable-flip-angle T1 mapping techniques are frequently applied in clinical settings, but their accuracy is often impaired by incomplete spoiling or flip angle miscalibrations. To eliminate these two error sources simultaneously, a combined B1 and T1 mapping method is proposed based on spiral 2D multislice spoiled gradient echo imaging with high spoiling efficiency. The transition to steady state is minimized by an optimized single preparation pulse. A single-shot spiral readout during the preparation module enables ultrafast B1 mapping and as a result reproducible bias-free T1 mapping with whole-brain coverage at clinically relevant resolution in less than one minute.

Introduction

Fast and accurate T1 quantification based on variable-flip-angle (VFA) spoiled-gradient-echo (SPGR) acquisitions is challenging due to the sensitivity of the acquired steady-state signal to transmit field inhomogeneities and incomplete spoiling. Here, a simultaneous spiral VFA B1 and T1 mapping method is presented that allows accurate whole-brain tissue T1 quantification in less than one minute a 1.5T and 3T.

Methods

MR acquisition scheme.

Interleaved 2D multislice SPGR acquisitions using a prototype sequence were performed with efficient long-TR and diffusion spoiling at two different nominal flip angles (α) 1: TR=250ms, total spoiler gradient area per TR (AG) = 459 (mT/m)∙ms, α1=17°, and α2=80° (cf. Fig. 1). The k-space was sampled with a 2D spiral-out trajectory (receive bandwidth: 400kHz) 2. Ten spiral interleaves were acquired at 1.5T with a readout time of 10.7ms. At 3T, the number of interleaves was increased to 20 and the readout time consequently shortened to 6.6ms to reduce off-resonance sensitivity. A short scan time of 2.5s and 5s was yielded at 1.5T and 3T, respectively, to acquire one slice group (11 slices, in-plane resolution: 1.3x1.3mm2, slice thickness: 3mm).

The transition to steady state was minimized by a single preparation pulse with optimized nominal flip angle β and recovery period Trec (cf. Fig. 1). The following parameters were used 1: β1=34° / Trec,1=250ms (for α1=17°) and β2=102° / Trec,2=350ms (for α2=80°). The free induction decay (FID) signal during the preparation module was sampled with an in-plane resolution of 5.3x5.3mm2 by a single-shot spiral acquisition (readout time: 12.4ms / 12.1ms at 1.5T / 3T).

Five slice groups were measured with interleaved positioning to obtain whole-brain coverage (i.e., in total 55 slices, overall scan time: 28s / 53s at 1.5T / 3T). Obtained B1 and T1 values were compared to standard reference methods. The geometry of the reference B1 acquisition (duration: 12s) 3 was matched to one of the five acquired slice groups.

B1 and T1 calculation.

The distribution of the transverse magnetization across the slice was calculated for N discrete samples based on Bloch simulations. An actual flip angle at the center of the slice profile of βact = cB1·βnom / αact = cB1·αnom was assumed for the preparation and excitation pulses (cB1: B1 scaling factor, βnom / αnom: nominal flip angles). The x and y components of the FID sampled during the preparation module can be described as $$$S_{\beta,x}=K\cdot\sum_{i=1}^N\sin\beta_{i}\sin\theta_{i}$$$ and $$$S_{\beta,y}=K\cdot\sum_{i=1}^N\sin\beta_{i}\cos\theta_{i}$$$, respectively, yielding a total signal amplitude of $$$S_{\beta}=\sqrt{S_{\beta,x}^2+S_{\beta,y}^2}$$$ (K: factor including proton density, coil sensitivity, T2* relaxation; βi / θi: flip angle / phase at the location i across the slice calculated for βact). An estimate for B1 (cB1) was derived by numerical minimization based on the signal ratio $$$S_{\beta_{1}}/S_{\beta_{2}}$$$. The raw B1 maps were median filtered. Similarly, T1 was derived based on the signal ratio $$$S_{\alpha_{1}}/S_{\alpha_{2}}$$$ where $$$S_{\alpha_{1}}$$$ and $$$S_{\alpha_{2}}$$$ are ideally spoiled steady-state signal amplitudes calculated as the summation over the slice profile 4 with B1-corrected flip angles.

Results

Excellent agreement with the reference B1 was observed both in vitro (0.125mM MnCl2 in H2O, mimicking tissue T1 and T2 conditions) and in vivo (human brain) as reflected by high intraclass correlation coefficients (ICC): 0.986 (1.5T) / 0.992 (3T) and 0.980 (1.5T) / 0.987 (3T) for the in vitro and in vivo scans, respectively (cf. Fig. 2). The peaks of the in vitro T1 histograms agree well with the reference T1 (cf. Fig. 3a), while the corresponding histograms of the same data are highly degraded if B1 inhomogeneities are not corrected (cf. Fig. 3b). The B1-corrected whole-brain T1 histograms obtained in a healthy volunteer at 1.5T and 3T show clearly delineated WM and GM T1 peaks in good agreement with literature values (cf. Ref. 5 and Fig. 3c). The corresponding B1 and T1 maps appear artifact-free (cf. Fig. 4).

Discussion and Conclusion

Reliable whole-brain spiral VFA T1 mapping of human brain tissues is achieved in less than one minute at both 1.5T and 3T. No additional scan time needs to be invested for B1 mapping since flip angle miscalibrations can be estimated in an ultrafast manner based on intrinsically coregistered contrasts acquired during the preparation module (cf. Fig. 1). In conclusion, the fast acquisition, volumetric coverage, and high accuracy accentuate the potential of the presented technique for clinical applications.

Acknowledgements

This work was supported by a grant from the Swiss National Science Foundation (SNF 325230-153332).

References

1. Heule R, Pfeuffer J, Bieri O. Snapshot whole-brain T1 relaxometry using steady-state prepared spiral multislice variable flip angle imaging. Magn Reson Med 2017. doi: 10.1002/mrm.26746.

2. Meyer CH, Zhao L, Lustig M, Jilwan-Nicolas M, Wintermark M, Mugler JP, Epstein FH. Dual-density and parallel spiral ASL for motion artifact reduction. Proceedings ISMRM 2011. p 3986.

3. Chung S, Kim D, Breton E, Axel L. Rapid B1+ mapping using a preconditioning RF pulse with TurboFLASH readout. Magn Reson Med 2010;64(2):439-446.

4. Parker GJ, Barker GJ, Tofts PS. Accurate multislice gradient echo T1 measurement in the presence of non-ideal RF pulse shape and RF field nonuniformity. Magn Reson Med 2001;45(5):838-845.

5. Rooney WD, Johnson G, Li X, Cohen ER, Kim SG, Ugurbil K, Springer CS, Jr. Magnetic field and tissue dependencies of human brain longitudinal 1H2O relaxation in vivo. Magn Reson Med 2007;57(2):308-318.

Figures

Figure 1. In the preparation module, the FID is acquired by a single-shot spiral readout. Preparation pulse flip angle β and recovery period Trec are optimized depending on the parameters of the imaging module (TR / flip angle α) 1. After the recovery period Trec (≥TR), the actual spiral 2D multislice SPGR acquisition starts, consisting of the interleaved excitation of 11 slices within each TR. The slices are separated by an edge-to-edge spacing of four times the slice thickness to mitigate slice cross-talk and magnetization-transfer-related issues. The spoiler gradient moments associated with the individual slices are illustrated as gray boxes.

Figure 2. Correlation scatter plots between the B1 scaling factor cB1 obtained with spiral VFA (x-axis) and the reference method (y-axis) for representative in vitro (a) and in vivo (b) measurements at 1.5T (left) and 3T (right). The agreement between the two techniques is assessed by linear fitting of the data (solid black line) and by the calculation of the intraclass correlation coefficient (ICC). The high ICC values and slope values close to 1 indicate excellent agreement.

Figure 3. In vitro (a/b) and in vivo whole-brain (c) T1 histograms obtained at 1.5T (left) and 3T (right). The histogram peak values are computed based on probability density estimation using the normal distribution as kernel function. (a/b) The location of the reference T1 value derived from single-slice inversion recovery turbo-spin-echo scans with inversion times of TI=[100, 200, 400, 800, 1600, 3200]ms is indicated by the orange line. (a/c) Flip angle miscalibrations are corrected by simultaneous B1 determination based on a single-shot spiral readout during the preparation module. (b) The data shown in (a) without correction of flip angle miscalibrations.

Figure 4. Representative in vivo B1 (here, cB1 = αact/αnom, i.e., actual/nominal flip angle) and T1 maps obtained in the brain of a healthy volunteer at both 1.5T (left) and 3T (right) with spiral 2D multislice VFA acquisitions. The same scaling is used for the 1.5T and 3T data to show the field dependence of B1 and T1. Note that the 1.5T and 3T measurements are not registered onto each other but similar slices are selected.

Proc. Intl. Soc. Mag. Reson. Med. 26 (2018)
0262