4971

Observation and Mitigation of Magnetization Transfer Effects in the two-point 3D Variable Flip Angle T1 Mapping Technique at 3T

Jean-David Jutras¹, Atefeh Kordzadeh², and Nicola De Zanche^3,4

¹Medical Physics, Cancer Centre of Southeastern Ontario, Kingston, ON, Canada, ²Biomedical Engineering, University of Alberta, Edmonton, AB, Canada, ³Oncology, University of Alberta, Edmonton, AB, Canada, ⁴Medical Physics, Cross Cancer Institute, Edmonton, AB, Canada

Synopsis

The Variable Flip Angle (VFA) T₁ mapping technique has been employed extensively in the past given its high contrast-to-noise ratio per unit scan time. However, its sensitivity to B₁ field inhomogeneity and imperfect spoiling hinders its reproducibility among different scanners and imaging centers. In this study, we investigate the impact of magnetization transfer (MT) effects in a 3D VFA technique using fixed flip angles, while varying B₁ amplitudes and durations, following corrections for non-ideal RF spoiling and RF inhomogeneity. We show that via careful tuning of the RF pulse amplitudes and durations, MT effects can be mitigated, yielding T₁ measurements that match closely with a gold-standard IR-EPI technique.

Introduction

Recently, the impact of on-resonance magnetization transfer has been investigated in several T₁ mapping techniques (e.g., MP2RAGE and IR-FSE [1]), to improve their reproducibility and accuracy [2]. Ou et al [3] reported MT effects in 2D (slice-selective) variable flip angle (VFA) T₁ quantification, which resulted in a systematic underestimation of the true T₁. In the 3D (non-selective) two-point VFA technique, some authors have obtained accurate T₁ maps (that agreed closely with gold-standard inversion recovery) after correcting for both flip-angle inhomogeneity and non-ideal RF spoiling [4], while others have observed a remaining yet unknown source of bias [2], [5]. In this study, we investigate the impact of MT effects (which we suspect to be this remaining source of bias) in a 3D VFA technique using fixed flip angles, while varying B₁ amplitudes and durations, following corrections for non-ideal RF spoiling and RF inhomogeneity.

Methods

Four healthy volunteers (ages 20-30) were scanned (after informed consent and ethics approval) on a Philips Achieva 3T scanner using 3D 9-echo (bipolar) VFA protocols (see Table 1). The T₁ was solved from the slope of the linearized Ernst equation:

$$ \frac{S_{ME}}{sin⁡(c_{RF}^{+} α_{nom} )} =E_{1} \frac{S_{ME}}{tan⁡(c_{RF}^{+} α_{nom} )}+c_{RF}^{-} M_0 (1-E_1 ) \sqrt{\sum_{n=1}^9e^{-2TE_n/T_2^*}}$$,

where S_ME is the combined signal (root-sum-of-squares) of the 9 echoes, c_RF⁺ is the ratio of the local to nominal flip angle (α_nom), E₁=exp(-TR/T₁), M₀ is the proton density and c_RF^- is the receiver coil profile. From this equation, it can be shown that the T₁^app (T₁solved by assuming c_RF⁺=1) is related to the true T₁ by $$$T_1^{app}\approx (c_{RF}^+)^2 T_1$$$. All scan parameters were kept identical in three versions of the sequence except that the duration and amplitude of the rectangular (non-selective) RF pulses were varied. Correction for the non-ideal RF spoiling was performed using the technique of Preibisch and Deichmann [5]. To guarantee the accuracy of the B₁ maps, two independent B₁ mapping sequences were acquired for comparison: Actual Flip angle Imaging (AFI) [6] and the Saturated Double Angle Method (SDAM) [7].The B₁ field was also simulated using Ansys HFSS on a rectangular phantom (with two different NaCl concentrations) placed inside a T/R birdcage coil and compared to phantom VFA, AFI and SDAM measurements. To compare the in vivo VFA T₁ with inversion recovery, a single-slice (encompassing important brain ROIs) 8-point IR-EPI sequence was also tested on the 4 subjects (using the T₁ mapping toolbox of Barral et al [8]). The T₁ maps from the 3 VFA protocols were co-registered and compared with IR-EPI at TI_min=20/200ms (including/avoiding MT effects).

Results

As shown in Figure 1, both AFI and SDAM yield a measured B₁ field very comparable to the HFSS simulations, while the common assumption of $$$T_1^{app}\approx (c_{RF}^+)^2 T_1$$$ (without a spoiling correction) fails if using a phase-cycle increment of ϕ₀=50°, especially in the low-loss phantom. Figure 2 demonstrates the significant impact of the different types of corrections on a whole-brain T₁ histogram, while Figure 3 shows the axial-slice comparison of IR-EPI at TI_min=20/200ms with VFA-SDAM at three different B₁ amplitudes. The ROI T₁ measurements in various brain regions (averaged across the 4 subjects) are tabulated for the different techniques in Table 2, demonstrating the close agreement (<0.5% difference between AFI and SDAM). As expected, varying the VFA B₁ amplitude had no impact on the T₁ map of a water phantom (<2% difference, not shown), where a single pool of liquid protons exists.

Discussion

The impact of on-resonance magnetization effects on T₁ in the VFA technique depends largely on the RF pulse duration and amplitude of the higher SPGR flip angle. Using high B₁ amplitudes yields higher T₁ (mean WM T₁≈975ms) than IR-EPI at TI_min=200ms, while low B₁ amplitudes yield significantly lower T₁ (WM T₁≈830ms), which is consistent with the results of Ou et al [3]. Rioux et al [1] recommended using TI_min>200ms at 3T to remove the shorter-decaying T₁ component. Our results demonstrate that by adjusting the B₁ amplitude and duration of the longer SPGR flip angle, VFA T₁ measurements can be made to match closely with IR-EPI mono-exponential T₁ in the brain.

Conclusion

In this work, we have demonstrated that, in addition to B₁ inhomogeneity and/or non-ideal RF spoiling, MT effects can also bias the resulting T₁, depending on the RF durations (and, obviously, the flip angles). The three sources of bias explain the broad range in mean WM T₁ reported at 3T: (~900ms [4] to ~1085ms [9], reproduced here in Figure 2) and must therefore be corrected. For better VFA T₁ accuracy/reproducibility in the future, the SPGR pulse durations and flip angles should be specified and fine-tuned (e.g. B₁≈6μT given α₁/α₂/TR≈5°/29°/24ms) to match the T₁ obtained with IR-EPI (with TI_min≥200ms).

Acknowledgements

The authors wish to thank Dr. Keith Wachowicz for accelerating the MATLAB simulation code used to obtain the spoiling correction coefficients. We also thank the Alberta Cancer Foundation for funding, Philips Healthcare for technical support, Dr. Roger Luechinger for the PATI program used for data transfer, as well as all volunteers who took part in this study.

References

[1] J. A. Rioux, I. R. Levesque, and B. K. Rutt, “Biexponential longitudinal relaxation in white matter: Characterization and impact on T1 mapping with IR-FSE and MP2RAGE,” (2016) Magn. Reson. Med.(75):2265–2277.

[2] N. Stikov, M. Boudreau, I. R. Levesque, C. L. Tardif, J. K. Barral, and G. B. Pike, “On the accuracy of T1 mapping: Searching for common ground,” (2015) Magn. Reson. Med. (73):514–522.

[3] X. Ou and D. F. Gochberg, “MT effects and T1 quantification in single-slice spoiled gradient echo imaging,”(2008) Magn. Reson. Med. (59):835–845.

[4] R. Heule, C. Ganter, and O. Bieri, “Variable flip angle T1 mapping in the human brain with reduced T2 sensitivity using fast radiofrequency-spoiled gradient echo imaging,” (2016) Magn. Reson. Med. (75):1413–1422.

[5] C. Preibisch and R. Deichmann, “Influence of RF spoiling on the stability and accuracy of T1 mapping based on spoiled FLASH with varying flip angles.,” (2009) Magn. Reson. Med., (61):125–135.

[6] V. L. Yarnykh, “Optimal radiofrequency and gradient spoiling for improved accuracy of T1 and B1 measurements using fast steady-state techniques.,” (2010) Magn. Reson. Med. (63):1610–1626.

[7] C. H. Cunningham, J. M. Pauly, and K. S. Nayak, “Saturated double-angle method for rapid B1+ mapping.,” (2006) Magn. Reson. Med.,(55):1326–1333.

[8] J. K. Barral, E. Gudmundson, N. Stikov, M. Etezadi-Amoli, P. Stoica, and D. G. Nishimura, “A robust methodology for in vivo T1 mapping,” (2010) Magn. Reson. Med. (64):1057–1067.

[9] H-L. M. Cheng and G.A. Wright. “Rapid high-resolution T1 mapping by variable flip angles: accurate and precise measurments in the presence of radiofrequency field inhomogeneity. (2006) Magn. Reson. Med.(55):566-574.

Figures

Table 1: MRI sequence parameters used. Note that the two SPGR sequences were repeated using identical flip angles, but different B₁ amplitudes of 1.0/2.0, 2.3/5.8 and 13.5/13.5µT (default) for α₁/α₂, respectively. For AFI, strong gradient spoiling was achieved with spoiler gradient areas of A_G1/A_G2=425/910 mT·ms/m. The IR-EPI sequence was repeated with the following inversion times: TI=20, 200, 500, 800, 1200, 1700, 2300 and 3000ms.

Table 2: Mean T₁ values (with standard deviations) measured in ROIs averaged across four healthy subjects for the two B₁ mapping techniques (AFI and SDAM). For the putamen, globus pallidus, caudate, frontal WM and occipital WM, the ROIs on the left-end and right-end sides of the brain were averaged.

Figure 1: Images (sagittal, coronal, and axial slices, respectively) of the measured and simulated c_RF⁺ maps in phantom 1 (lossy) and 2 (low-loss), along with corresponding profile plots (red, blue and green panels). The expected c_RF⁺ from the assumption T₁^app≈ (c_RF⁺)² T₁ is obtained from the T₁^app of a weakly-spoiled multi-echo SPGR (B₁=13.5μT) and by assuming a uniform T₁(=1353/1471 ms for lossy/low-loss phantoms measured using IR-EPI). Phantom (dimensions: 25.0×15.2×14.0cm³) compositions are: 4.5L of 1% agar solution, 0.05%/0.05% NaN₃, 35μM/31μM MnCl₂ and 0.2%/0% NaCl, yielding different conductivities but T₁ and T₂ comparable to GM for the lossy/low-loss, respectively.

Figure 2: Whole-brain T₁ histograms (subject 1) at high B₁ amplitude shown without any correction (red), with the B₁⁺ correction using SDAM (black), with both B₁⁺ and RF spoiling correction (green), and with all three corrections included (magenta, where MT effects are minimized using the medium B₁ pulses).

Figure 3: Comparison of IR-EPI (at two different TI_min) with VFA-SDAM at low, medium and high B₁ amplitude (subject 1). Observe how the IR-EPI with MT effects (TI_min=20ms) compares with the VFA-SDAM image at low B₁ (in the WM), while the IR-EPI without MT effects (TI_min=200ms) compares with the VFA-SDAM at medium B₁ amplitude. The VFA-SDAM image at high B₁ clearly yields longest T₁.

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

4971