Analyzing Myelin Water Fraction using mcRISE
Fang Liu1, Andrew Alexander2, and Alexey Samsonov1

1Department of Radiology, University of Wisconsin-Madison, Madison, WI, United States, 2Department of Medical Physics, University of Wisconsin-Madison, Madison, WI, United States

Synopsis

Myelin, a thin layer of sheath-like cell, provides an important role in protecting nerve axon and accelerating neural impulse transmission. Myelin water fraction (MWF) mapping has been recently proposed for assessing myelin content in-vivo. One quantitative MR method called mcDESPOT has shown promising results for assessing myelin content. However, this method is lack of consideration of magnetization transfer (MT) effect leading to the complication of interpretation for MWF values. In this study, we proposed a method called mcRISE to account for MT effect and investigate the feasibility of assessing myelin content with MT-insensitive MWF as well as additional MT parameters.

Introduction

Myelin, a thin layer of sheath-like cell, provides an important role in protecting nerve axon and accelerating neural impulse transmission. Recently a quantitative MR method called multicomponent-driven equilibrium single-component observation of T1 and T2 (mcDESPOT) has shown promising results for assessing myelin content using rapid steady-state sequences (1). In this method, a bicomponent water exchange system is used to characterize myelin water and non-myelin water. Yet this simplified bicomponent model is practically useful, the lack of consideration of interaction between non-aqueous macromolecular proton and mobile water through magnetization transfer (MT) renders the complication of interpretation for MWF values (2-4). Recent studies show that the steady-state sequence signal could be strongly biased by MT under certain imaging conditions (5,6). In our previous study, we proposed a new method called multicomponent relaxation imaging using steady-state signal evolution (mcRISE) in cartilage imaging to account for MT effect while preserve the advantageous acquisition of mcDESPOT (7). In this study, we optimized and applied the mcRISE method to the neural tissue and investigate the feasibility of assessing myelin content with MT-insensitive MWF as well as additional MT parameters.

Theory

In mcRISE, a macromolecular proton pool in exchange with mcDESPOT water pools was introduced to explicitly model MT exchange between macromolecular proton, myelin and non-myelin water. A combined mcDESPOT and quantitative magnetization transfer imaging (qMTI) (8) method is proposed to resolve this three compartmental system. Simultaneous fitting for all parameters of this system is not feasible. Hence, an incremental fitting approach was introduced. First, the macromolecular bound pool fraction f and exchange rate k was estimated using balanced steady-state free precession (bSSFP) based estimation of the MT parameters (8), where varying RF pulse width (bSSFPv) was applied for exploring on-resonance MT effect. Then, these values as fixed parameters were implemented in the full model fit with additional varying flip angle spoiled gradient echo (SPGR) and bSSFP, which involves the mutually exchanging slow and rapid relaxing water compartments and the bound proton pool. The detailed processing pipeline is shown in Figure 1.

Methods

A brain scan was performed on a healthy adult volunteer using a GE MR750 3.0T scanner. The mcRISE parameters included: 1) SPGR scans with TR/TE=4.1/1.5ms over a range of flip angles (α=3, 4, 5, 6, 7, 9, 13, 18°); 2) Two bSSFP scans with RF phase cycling on (bSSFP180) and off (bSSFP0), with TR/TE=3.0/1.4ms over a range of flip angles (α=2, 5, 10, 15, 20, 30, 40, 50°); 3) One bSSFPv scans over a range of RF pulse width (TRF) at TRF/TR=0.27/2.5ms, 0.3/2.5ms, 0.4/2.6ms, 0.6/2.8ms, 0.8/3.0ms, 1.2/3.4ms, 1.6/3.8ms, 2/4.2ms and α=35°. 4) Inversion recovery SPGR scan with TR/TE=4.1/1.5ms, TI=450ms, and α=5°. All scans were performed in the sagittal plane with a 25.6cm field of view, 128×128 matrix, 2mm thickness, one excitation.

Results & Discussion

mcRISE was able to simultaneously generate MWF map and additional quantitative MT f map of the entire brain at 3.0T (Figure 2). The spatial variations of the MWF and f maps within the brain tissue are consistent with the previous reported values from multiple multicomponent T2 and quantitative MT studies (1,8). A significant reduced MWF values is observed in the mcRISE method compared to mcDESPOT. A recent study comparing the MWF between mcDESPOT and a multi-echo T2 decay method demonstrated an overestimation of mcDESPOT MWF values in brain tissue, which was largely attributed to MT effects (3). The MT-corrected MWF values from mcRISE shows a better correspondence to reported multiecho T2 MWF and our previous simulation results (7). In addition, a strong correlation (p<0.0001 in Pearson test) between MWF and macromolecular proton fraction f for both mcRISE and mcDESPOT is observed (Figure 3) which corresponds well to one previous study comparing mcDESPOT MWF and f obtained from traditional off-resonance MT method (9), indicating the strong correlation between myelin content and macromolecular proton from qMT. Our previous in-vivo qMT studies also show the primary source of qMT f comes from the myelin sheath (10). However, this correlation is decreased for mcRISE (r2=0.41) as compared to mcDESPOT (r2=0.78), indicate that the MT corrected MWF might preserve complimentary information from that of MT measurements, although detailed studies comparing MWF and qMT parameters in terms of sensitivity and specificity to myelin content and structure are still required. In conclusion, mcRISE is capable of simultaneously estimating MT corrected MWF and qMT parameters in whole brain, combination of which might provide a new set of biomarkers for rapid and comprehensive assessing myelin content and structure in multiple brain studies.

Acknowledgements

The work was supported by NIH R01NS065034, NIH R01AR068373-01, and GE Healthcare.

References

1. Deoni SC, Rutt BK, Arun T, Pierpaoli C, Jones DK. Gleaning multicomponent T1 and T2 information from steady-state imaging data. Magn Reson Med 2008;60(6):1372-1387.

2. Zhang J, Kolind SH, Laule C, MacKay AL. How does magnetization transfer influence mcDESPOT results? Magn Reson Med 2014.

3. Zhang J, Kolind SH, Laule C, Mackay AL. Comparison of myelin water fraction from multiecho T2 decay curve and steady-state methods. Magn Reson Med 2014.

4. Lenz C, Klarhöfer M, Scheffler K. Limitations of rapid myelin water quantification using 3D bSSFP. MAGMA 2010;23(3):139-151.

5. Crooijmans HJ, Gloor M, Bieri O, Scheffler K. Influence of MT effects on T(2) quantification with 3D balanced steady-state free precession imaging. Magn Reson Med 2011;65(1):195-201.

6. Bieri O, Scheffler K. On the origin of apparent low tissue signals in balanced SSFP. Magn Reson Med 2006;56(5):1067-1074.

7. Liu F, Block WF, Kijowski R, Samsonov A. Rapid multicomponent relaxometry in steady state with correction of magnetization transfer effects. Magn Reson Med 2015.

8. Gloor M, Scheffler K, Bieri O. Quantitative magnetization transfer imaging using balanced SSFP. Magn Reson Med 2008;60(3):691-700.

9. Samson RS, Deoni SC, Wheeler-Kingshott CA. Correlation of potential myelin measures from quantitative Magnetisation Transfer (qMT) and multi-component Driven Equilibrium Single Pulse Observation of T1 and T2 (mcDESPOT). Proc Intl Soc Mag Reson Med. Hawaii, USA2009, abstract 4476.

10. Samsonov A, Alexander AL, Mossahebi P, Wu YC, Duncan ID, Field AS. Quantitative MR imaging of two-pool magnetization transfer model parameters in myelin mutant shaking pup. Neuroimage 2012;62(3):1390-1398.

Figures

Figure1. mcRISE processing workflow

Figure 2, MWF map obtained from mcDESPOT and mcRISE and f map obtained from mcRISE for one reformatted axial slice of healthy volunteer is shown in a. The corresponding histograms in b shows the significant overestimated MWF in mcDESPOT compared to mcRISE

Figure 3, Scatter plots for correlating voxel-wise MWF values obtained from mcDESPOT and mcRISE, and f values obtained from mcRISE. Both estimated MWF values has high correlation (p < 0.0001) with f values in the brain tissue in Pearson correlation analysis. However, MWF obtained from mcRISE has reduced linear correlation of 0.41 compared to mcDESPOT of 0.78.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1275