0787

Towards short dipolar relaxation time, T1D, MRI

Gopal Varma¹, Patricia Coutinho de Souza¹, Valentin H Prevost², Olivier M Girard², Victor Carvalho², Samira Mchinda², Leo Tsai¹, Guillaume Duhamel², Aaron K Grant¹, and David C Alsop¹

¹Division of MR Research, Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, United States, ²CNRS, CRMBM, UMR 7339, Aix Marseille Universite, Marseille, France

Synopsis

The inhomogeneous magnetization transfer (ihMT) technique has shown myelin sensitivity, and is understood to be dependent on power and the dipolar relaxation time parameter, T_1D, which is longer in myelinated tissues. Implementation of ihMT can be adapted to provide a smaller, but non-negligible signal from other, relatively short T_1D tissues. Simulations showed a measurable ihMT signal, achieved from fixed low duty cycle MT preparations with high B₁ pulses, decayed with pulse width at a rate dependent on T_1D. Thus short, high B₁ pulses were implemented to acquire ihMT data from ex-vivo samples of rat heart, kidney, and tail tendon, demonstrating the feasibility of short T_1D imaging.

Introduction

The inhomogeneous magnetization transfer (ihMT) technique gained attention for its myelin-associated contrast.¹ Further understanding of the nature of the ihMT signal associated its dependence on residual dipolar coupling.² In terms of the two-pool model of MT, extended using spin temperature formalism, the ihMT signal is mediated by the dipolar relaxation time, T_1D, parameter.^3,4 T_1Ds measured in ex-vivo white matter (WM) and gray matter (GM) are relatively long (at physiological temperature) compared to other tissues, and explain the selective signal observed in vivo.⁵ However the correlation of ihMT with T_1D also provides evidence that an ihMT signal (albeit small) is measurable in other tissues.³ The use of low duty cycle ihMT preparations shows a measurable ihMT signal from the relatively short T_1D tissue of muscle.⁶ Manning et al. have also shown that an ihMT signal is measurable from other non-lipid-based samples and point out its dependence on the power T_1D product.⁴ We explore the use of short (of varying width), high B₁ pulses in a low duty cycle preparation to obtain ihMT signal from organs outside the central nervous system to assess the feasibility of ihMT MRI outside the brain.

Methods

Simulations were conducted based on steady-state solutions of the two-pool MT model differential equations for pulsed MT experiments⁷, with and without terms for the dipolar order of the restricted pool to account for single and dual frequency MT preparations.^3,8 WM and GM parameters were taken from high-field quantitative MT literature.⁹ Tissues were excised from an 8-week old female Sprague-Dawley rat, in accordance with institutional animal care and use committee protocol, and placed in an abundance of formalin for ≥48hours, before placement in smaller containers for MRI. MRI was conducted on a 9.4T scanner (Biospec, Bruker) with a volume transmit coil and 4-channel receive mouse brain array. Temperature was monitored using a fiber optic probe (±1°C) and regulated by heated forced air (SA Instruments). IhMT data were acquired with single or dual off-resonance frequency preparatory pulses (Gaussian, with 0.4 shape ratio compared to rectangular pulse of same duration) applied every 50ms for 1s (B_1,RMS=3μT), followed by 48 gradient-echo readouts (TE/TR=2.8/6.7ms), repeated to cover all phase encodes required for a 3D volume. Off-resonance RF was applied at: ±7kHz in the brain; ±10kHz for other tissues, and dual frequency preparations were achieved by appropriate cosine modulation. The ihMT signal was calculated as the difference between data prepared with single or dual frequency RF irradiation, and division by the signal following a B_1,RMS=0μT preparation provided an ihMT ratio (ihMTR).

Results

The simulated ihMTRs for WM and GM decayed with pulse width for a fixed repetition period and B_1,RMS over that period (Fig. 1). A faster decay is apparent for shorter simulated T_1D values, whereas for changes in the simulated exchange rate the decay rate remained more consistent. Differentiation of deeper cerebellum (Dc) from surrounding cell layer (Cl) was improved at 36°C (Fig. 2). Short 1ms pulses still provided a visible ihMTR contrast between WM and GM regions at room temperature (Figs. 2-3). Preparation with such short, high B₁ pulses provided average ihMTRs ≥20% in the heart, renal cortex, and tail tendons (Fig. 4). The ihMTR in the annulus was twice that in the nucleus of a tail intervertebral disc (IVD). All samples showed the decrease in ihMTR with pulse width expected from simulations, while MTR (from dual frequency data) increased slightly (Fig. 5).

Discussion

Simulation of ihMTR as a function of pulse width for a fixed repetition period in the ihMT preparation suggests a decay that is more rapid for decreasing T_1D values (Fig. 1). However the simulated ihMTRs suggest a convergence for an infinitesimal pulse width. Thus short, high B₁ pulses should provide ample ihMT signal in relatively short T_1D tissues, and implementation of 1ms Gaussian pulses every 50ms with B_1,RMS=3μT was shown to do so for ex-vivo samples of rat: brain at 24°C, heart, kidney, IVD and tendons (Figs. 2b,4). The difference in ihMTR for 1ms pulses and the decay with pulse width at 36°C and 24°C, in various regions of the brain, suggests a change in exchange rate (Figs. 1c-d,3). Relatively slow decay in ihMTR compared with simulation was likely the result of a mismatch in pulse shapes (Figs. 1a-b,5). Given the expectation of T_1D mediated decay, further work will explore the potential for ihMT signal variation with pulse width to assist with quantification of T_1D.

Conclusions

The use of a low duty cycle preparation, i.e. short, high B₁ pulses, provided measurable ihMT signal in ex-vivo tissues outside of the brain. The simulations and results provide a framework for translation in-vivo to explore the utility of ihMT MRI of relatively short T_1D tissues.

Acknowledgements

No acknowledgement found.

References

1. Varma G, Duhamel G, de Bazelaire C, et al. Magnetization transfer from inhomogeneously broadened lines: A potential marker for myelin. Magn Reson Med. 2015;73(2):614-622.

2. Girard OM, Prevost VH, Varma G, et al. Magnetization from inhomogeneously broadened lines (ihMT): Experimental optimization of saturation parameters for human brain imaging at 1.5 Tesla. Magn Reson Med. 2015;73(6):699-709.

3. Varma G, Girard OM, Prevost VH, et al. Interpretation of magnetization transfer from inhomogeneously broadened lines (ihMT) in tissues as a dipolar order effect within motion restricted molecules. J Magn Reson. 2015;260:67-76.

4. Manning AP, Chang KL, MacKay AL, et al. The physical mechanism of “inhomogeneous” magnetization transfer MRI. J Magn Reson. 2017;274:125-136.

5. Swanson SD, Malyarenko DI, Fabiilli ML, et al. Molecular, dynamic, and structural origin of inhomogeneous magnetization transfer in lipid membranes. Magn Reson Med. 2017;77(3):1318-1328.

6. Varma G, Grant A, Girard O, et al. Modeling the increased inhomogeneous magnetization transfer (ihMT) signal from high amplitude, low duty cycle irradiation. Proc ISMRM. 2017:1986.

7. Portnoy S and Stanisz GJ. Modeling pulsed magnetization transfer. Magn Reson Med. 2007:58(1):144-155.

8. Sled J and Pike G. Quantitative interpretation of magnetization transfer in spoiled gradient echo sequences. J Magn Reson. 2000:145;24-36.

9. Naumova AV, Akulov AE, Khodanovich MY, et al. High-resolution three-dimensional macromolecular proton fraction mapping for quantitative neuroanatomical imaging of the rodent brain in ultra-high magnetic fields. NeuroImage 2017;147:985-993.

Figures

Simulated ihMTRs using average (a,c) WM and (b,d) GM parameter values taken from literature as a function of the widths of rectangular MT pulses applied repeatedly every 50ms. a-b) The ihMT signal was simulated for successively halved T_1D values to show the effect on the curve, as well as for, c-d) a fixed T_1D=3.0ms and exchange rate values, R doubled (dashed lines) and halved (dotted lines) relative to those provided in literature.

Sample ihMTR maps calculated from axially acquired data through the ex-vivo rat brain close to, a) physiological, and b) room temperature. IhMT data (two datasets with single and dual frequency preparations each, as well as two with B_1,RMS=0μT) were acquired with the following parameters: FA=8°; FOV=28.8x28.8x19.2mm3; matrix=96x96x32; centric phase encoding and linear slice phase encoding, for a 1min24s scan time per 3D volume. The ihMTR maps were masked to zero pixels corresponding to ≤10% of the maximum signal in the B_1,RMS=0μT images.

Plot of ihMTR decay with RF pulse width, in different regions of the ex-vivo rat brain, at a temperature close to physiological (solid lines) compared with data acquired close to room temperature (dashed lines). Average ihMTRs were measured in the slices shown in Figure 2: Nc = neocortex; Cc = corpus callosum; Hf = hippocampal formation; Th = thalamus; Cp = corticofugal pathways. Error bars represent the standard error over the region of interest.

Data acquired in samples of ex-vivo rat: a) heart; b) kidney; c) tail intervertebral disc (IVD); and d) tail tendons, suspended in formalin at 36°C. The leftmost column shows images acquired following zero power preparations and used to normalize the ihMT signal and calculate the ihMTRs. Other columns show the corresponding maps of ihMTR for different MT pulse widths. Data were acquired with: FOV=19.2x19.2x12.8mm3; matrix=96x96x32; 1min24s scan time per 3D volume in the heart and tail, and FOV=28.8x28.8x9.6mm3; matrix=96x96x16; 42s scan time per 3D volume in the kidney. Double the ihMT data were acquired in the tail to improve signal-to-noise.

Plots of average a) ihMTR, and b) MTR, measured in regions of interest within ex-vivo rat tissues suspended in formalin at 36°C. Data acquired in a sample of hair conditioner (grey line) provides a relatively long T_1D reference to compare the decay of ihMTR as a function of the off-resonance pulses widths. MT pulses widths represent the full length, whereas the full-width at half-maximum for 1, 3, 5, and 7ms Gaussian pulses were 0.4, 1.2, 2.0, and 2.8ms respectively. MTR values were calculated from dual frequency MT preparation data. Error bars represent the standard error over each region of interest.

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

0787