In-vivo measurement of a new source of tissue contrast, the dipolar relaxation time,T1D, using a modified ihMT sequence
Gopal Varma1, Valentin H Prevost2, Olivier M Girard2, Guillaume Duhamel2, and David C Alsop1

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

Synopsis

The enhanced inhomogeneous magnetization transfer (ihMT) in certain tissues, especially white matter, has recently been explained as a result of longer dipolar relaxation times, T1Ds in those tissues. Measurement of T1D by modeling the frequency and power dependence of steady state ihMT has yielded T1D estimates but with great uncertainty. Here we introduce a dynamic ihMT experiment that switches between positive and negative frequency irradiation at varying times. Fits to the ihMT signal decay curve as a function of switching time at one (absolute) offset frequency and power enabled highly precise mapping of T1D that was largely independent of other MT parameters. A T1D of 6.4±0.5ms for white matter was in good agreement with reported ex-vivo measurements using Jeener-Broekaert echoes.

Purpose

To develop a fast and accurate method to measure the dipolar relaxation time, T1D, in-vivo.

Introduction

The contrast mechanism described as inhomogeneous magnetization transfer (ihMT) is based on the signal difference following saturation at a single offset frequency and that following dual frequency saturation [1-2]. This ihMT signal has been recently modeled based on the interaction between the bound and free pool, with and without inclusion of a dipolar component to the bound pool [3]. In the case of a pulsed saturation preparation, the ihMT signal is found to be dependent on the number of pulses between switching frequencies for the dual offset experiment [4]. In this work, the model that describes the ihMT signal was adapted to describe the effect of a variation in the period between switches in frequencies. In doing so, we provide a robust method by which a dipolar relaxation time, T1D can be extracted from ihMT data from different substances.

Methods

A modified ihMT pulse sequence was developed with MT saturation performed using 0.5ms Hann shaped RF pulses every 1.2ms. Dual frequency irradiation is typically achieved by alternating between positive and negative frequency pulses. By increasing the number of pulses performed at the same frequency before switching to the negative frequency, a frequency switching time, w was introduced (Fig.1). An approximate analytic solution to the dipolar ihMT model, described in [3], was derived to enable rapid fitting of the signal decay curves as a function of switching time, w. Data were acquired in 5 healthy volunteers and hair conditioner (hc) on a GE 3T scanner with 500ms of saturation at B1,RMS=3.5μT, |Δ|=7kHz, and 8 values for w ranging from 1.2-43.2ms, followed by a single-shot spin-echo EPI acquisition. ihMT ratios (ihMTRs) calculated from ROI analysis, e.g. of white matter (WM) and gray matter (GM) (Fig.2a), were fit to the model for ihMT(w). Similarly, ihMT(w) data acquired at 11.75T from WM, GM and muscle (mu) ROIs in 2 mice, 2/4% agarose (aga), and hc #2 with a short TE single-shot FSE sequence and 1ms RF pulses every 1.3ms at |Δ|=8kHz and w=1.3-42.9ms, were also fit to the model. Mouse in-vivo data were collected following saturation at B1,RMS=4.7;5.8μT for 0.75s; whilst for aga and hc 11.75T data, B1,RMS=6;8.5;12μT for 1.2s. Initial values for exchange/relaxation parameters were based on previous literature [3,5-6].

Results

Initial estimates for parameters other than dipolar relaxation time, T1D, i.e. R, RM0B/RA, 1/RAT2A, and T2B, showed large variation. Fixing these parameters to the initial values, the ihMT(w) model was modified to have two variable parameters: T1D and a constant, A by which the resultant ihMTR was multiplied, such that ihMTR=A*ihMT(w,T1D) (Fig.2b). T1D remained relatively stable despite factor of 2 changes in the fixed parameters: R, RM0B/RA, 1/RAT2A, and T2B. The T1D values found in GM were similar to that in WM for both humans and mice (Fig.3); A in human GM was 0.54 times that of WM in the posterior region (Fig.2c). T1Ds extracted from fitting were longer in samples associated with (previously reported) higher ihMTR values. Indeed, T1D was 100 times longer in hc #2 than in aga; hc #1 sample chilled to 3°C showed a 35% increase in T1D, compared to that at room temperature.

Discussion

Although there was variability in the amplitude and/or other parameters related to the model, the shape of the ihMT signal as a function of w (as defined for this type of experiment, Fig.1) was mostly dictated by the T1D (Fig.2b). As in previous works, measurements of T1D in WM/GM were longer than those in mu [3]. Further work is warranted, in particular detailed comparison with alternative measurements of T1D, e.g. using Jeener-Broekaert echoes [7-8].

Conclusions

The experiments and model outlined provided a means by which the dipolar relaxation time, T1D may easily be measured in-vivo (independent of other MT parameters). This approach may be used to improve modeling of ihMT and to quantitatively characterize pathologic tissues.

Acknowledgements

We would like to thank F. Kourtelidis for help with volunteer scans.

References

[1] Varma et al. Magn Reson Med 73:614-22 (2015). [2] Girard et al. Magn Reson Med 73:2111-21 (2015). [3] Varma et al. J Magn Reson 260:67-76 (2015). [4] Varma et al. Proc Intl Soc Mag Reson Med 21:2536 (2013). [5] Morrison et al. J Magn Reson B 108:103-13 (1995). [6] Henkelman et al. Magn Reson Med 29:759-66 (1993). [7] Jeener et al. Phys Rev 157:232-40 (1967). [8] Swanson et al. Proc Intl Soc Mag Reson Med 23:0994 (2015).

Figures

Figure 1 Illustration of pulsed MT experiment that provides dual frequency offset (Δ) saturation for a) w=1.2ms, and b) w=3.6ms.

Figure 2 a) Image of ihMT contrast (FOV=25x25cm2;matrix=128x128;slice=6mm) illustrating ROIs (frontal WM; ventricle adjacent WM; splenium of corpus callosum; posterior WM and GM) used for analysis. b) Plot of ihMTR(w) comparing fits to model with variable MT parameters, and two variable parameters. c) Results from two parameter fits to ROI data.

Figure 3 Table of results from 2 parameter fit of model to ihMT(w) data from ROIs within different tissues/samples. When possible, values given with standard error.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0802