4588

Multi-contrast Multi-resolution UTE for Myelin Fraction Mapping

Nan Yin¹, Shuai Liu¹, Marco Reisert¹, Serhat Ilbey², Alexander Rau³, Uzay Emir^4,5, Michael Bock¹, and Ali Caglar Özen¹
¹Division of Medical Physics, Department of Diagnostic and Interventional Radiology, University Medical Center Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, ²Bruker BioSpin, Karlsruhe, Germany, ³Deparment of Diagnostic and Interventional Radiology and Neuroradiology, University Medical Center Freiburg, Freiburg, Germany, ⁴Weldon School of Biomedical Engineering, Purdue University, West Lafayette, IN, United States, ⁵School of Health Sciences, College of Health and Human Sciences, Purdue University, West Lafayette, IN, United States

Synopsis

Keywords: Quantitative Imaging, Quantitative Imaging, Myelin, UTE, Bayesian

Motivation: The direct detection of myelin is not possible due to extremely short T2 of the myelin bilayer. Ultra-short T2* component can be calculated with ultra-short echo-time (UTE) sequences.

Goal(s): To evaluate the performance of the multi-contrast multi-resolution 3D-UTE sequence (mcUTE) in myelin fraction estimation.

Approach: Combining 1-1 binomial water-excitation pulses and zero-moment encoding in between, mcUTE enables simultaneous acquisition of quantitative and high-resolution water-excited structural images. Myelin fraction from mcUTE is compared to myelin water fraction method.

Results: Myelin fraction values were comparable with literature, and consistent within the subjects.

Impact: mcUTE can potentially be used as a single imaging modality for myelin fraction estimation since it produces both quantitative and a high resolution water-excited anatomical image data in 18mins with TE=20μs. Our results show the acquisition can be further accelerated.

Introduction

Multiple sclerosis is one of the most common neurologic diseases and a combination of neuroinflammation and degeneration leads to a reduced white matter myelin content which can be imaged with MRI. Conventional (i.e. qualitative) MRI techniques lack lesion specificity and sensitivity. In contrast, quantitative MRI (qMRI) yields additional information on the brain tissue and pathologies¹. In particular, myelin water fraction (MWF) imaging was realized using T1, T2 or T2* information^2–7. Magnetization transfer (MT) contrast^8,9 and diffusion-weighted MRI (DWI)¹⁰ were proposed as more direct indicator of myelin content. Recently, ultra-short time echo (UTE) sequences¹¹ were introduced as a potential biomarker of the myelin content. However, the reproducibility and specificity of qMRI is rather heterogeneous¹² and a recent meta-analysis reported only limited correlation with histology^13,14.

Previously, we used a 1-1 binomial water excitation hard pulse in a UTE sequence to achieve high-resolution structural imaging, and added a bipolar gradient pair to acquire multi-contrast multi-resolution UTE (mcUTE) data sets¹⁵. In this study, this sequence was used to measure the ultra-short T2* (us-T2*) signal components in the brain to calculate myelin fraction (MF) maps. We represented the MR signal as weighted sum of three relaxation components and defined the ratio of the weighting of the us-T2* component to total weights of all components as MF. The sequence was tested in three healthy volunteers, and MF maps were obtained using Bayesian learning. Finally, MFs were compared to conventional MWF methods and literature values.

Methods

In mcUTE¹⁵, a bipolar readout gradient that acquires two echoes (TE1, TE2, bandwidth=635Hz/px, resolution=1.71mm) was inserted into a 1-1 binomial water excitation hard pulses (Fig.1). Immediately after the second RF pulse, a third echo (TE3) was acquired with TE3=20µs, bandwidth=635Hz/px, and resolution=0.50mm. To map us-T2* components, the mcUTE acquisition was repeated 9 times with different TEs for the TE1 and TE2. In Fig.2, T1w-MP-RAGE (TR=2000ms, FOV=224mm, resolution=1mm, α=8°, bandwidth=220Hz/px) and T2w-SPACE (TR=2500ms, FOV=224mm, resolution=1mm, α=90°, bandwidth=685Hz/px) sequences were acquired with varied TIs (850,900,980,1050,1200ms) and TEs (22,44,98,131,178ms) to obtain quantitative T1 and T2 maps for MWF. A 3D FLASH sequence (TR/TE=50/5,10,20,30,45ms, FOV=224mm, resolution=1mm, α=90°, bandwidth=350Hz/px) was used to quantify long T2* components, since it is impracticable to assign TE > 10ms for mcUTE. All sequences were tested in 3 healthy volunteers on a 3T MRI system (Magnetom PRISMA, Siemens).

T2* and MF maps were computed voxel-wise from the signal intensity of different TEs obtained from mcUTE. A three-component model was used: $$$S(T E)=S_0 e^{-T E / T_{2, ultrashort}^*}+S_1 e^{-T E / T_{2, short }^*}+S_2 e^{-T E / T_{2, long }^*} $$$, where the long T2* information $$$T_{2, long }^* $$$ was obtained from a 3D FLASH acquisition. A two-component model without the long T2* value from 3D-FLASH was also tested, where the signal at TE=1160µs was subtracted from that of at the remaining TEs in mcUTE.

For model parameter estimation, we used a Bayesian estimator: $$$\tilde{x}_B(S)=\int x p(x \mid S) d x$$$, where x are the parameters ($$$ S_0, T_{2, ultrashort}^*, S_1, T_{2, short }^*, S_2 $$$) of the signal model. To find $$$\tilde{x}_B(S) $$$, we use a polynomial regressor to represent the mapping $$$\tilde{x}_B(S) $$$ and minimize the quadratic loss function $$$ L(\tilde{x}, x)=(\tilde{x}-x)^2$$$: $$$ \tilde{x}_B(S)=\underset{\tilde{x}}{\operatorname{argmin}} \int L(\tilde{x}(S), x) p(x, S) d x d S$$$, where the integral is computed by Monte Carlo methods¹⁶.

Results

In Fig. 3, exemplary data sets are shown for a 35y-old male volunteer. Quantitative brain images with a signal intensity gradually decaying and high-resolution water-excited structural image are depicted. MF and us-T2* for all three subjects were 5.3±2.7/7.4±2.0/8.3±0.8/6.7±0.9 % and 59.0±15.2/62.3±7.2/64.8±4.7/60.2±7.6 µs for GM/WM/Corona radiata/External capsule (Fig.4). In Fig.5, MF and T2* maps of the same axial slices from the same volunteer show that the myelin fraction in WM is in general homogeneous and about 50% higher than in GM, consistent with the results reflected by MWF (structural similarity: SSIM(MF,MWF)=0.784). Long-T2* are shorter for WM compared to GM, which is consistent with the literature¹⁷.

Discussion

In this study, long acquisition times are the main limitations. Acquisition of mcUTE sequences can be improved by incorporating additional echoes to estimate the long T2* components. Additional information such as B1 and T1 can also be estimated using the relationship between the first and the third echo. Variable flip angles, TR and TE can be integrated to obtain quantitative maps more rapidly similar to MR-Fingerprinting approach¹⁸. Comparison of the results with histology and other myelin imaging methods such as DWI and MT are planned to further our understanding of the utility of mcUTE.

Acknowledgements

No acknowledgement found.

References

1. Weiskopf N, Edwards LJ, Helms G, Mohammadi S, Kirilina E. Quantitative magnetic resonance imaging of brain anatomy and in vivo histology. Nat Rev Phys. 2021;3(8):570-588. doi:10.1038/s42254-021-00326-1

2. Mackay A, Whittall K, Adler J, Li D, Paty D, Graeb D. In vivo visualization of myelin water in brain by magnetic resonance. Magnetic Resonance in Medicine. 1994;31(6):673-677. doi:10.1002/mrm.1910310614

3. Alonso-Ortiz E, Levesque IR, Pike GB. MRI-based myelin water imaging: A technical review. Magn Reson Med. 2015;73(1):70-81. doi:10.1002/mrm.25198

4. Oh SH, Bilello M, Schindler M, Markowitz CE, Detre JA, Lee J. Direct visualization of short transverse relaxation time component (ViSTa). Neuroimage. 2013;83:485-492. doi:10.1016/j.neuroimage.2013.06.047

5. Lee J, Hyun JW, Lee J, et al. So You Want to Image Myelin Using MRI: An Overview and Practical Guide for Myelin Water Imaging. Journal of Magnetic Resonance Imaging. 2021;53(2):360-373. doi:10.1002/jmri.27059

6. MacKay AL, Laule C. Magnetic Resonance of Myelin Water: An in vivo Marker for Myelin. Brain Plast. 2(1):71-91. doi:10.3233/BPL-160033

7. Myelin water imaging to detect demyelination and remyelination and its validation in pathology - Laule - 2018 - Brain Pathology - Wiley Online Library. https://onlinelibrary.wiley.com/doi/10.1111/bpa.12645. Accessed November 3, 2023.

8. Duhamel G, Prevost VH, Cayre M, et al. Validating the sensitivity of inhomogeneous magnetization transfer (ihMT) MRI to myelin with fluorescence microscopy. Neuroimage. 2019;199:289-303. doi:10.1016/j.neuroimage.2019.05.061

9. Varma G, Girard OM, Prevost VH, Grant AK, Duhamel G, Alsop DC. 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. doi:10.1016/j.jmr.2015.08.024

10. Chan KS, Marques JP. Multi-compartment relaxometry and diffusion informed myelin water imaging – Promises and challenges of new gradient echo myelin water imaging methods. NeuroImage. 2020;221:117159. doi:10.1016/j.neuroimage.2020.117159

11. Ma YJ, Jang H, Lombardi AF, Corey-Bloom J, Bydder GM. Myelin water imaging using a short-TR adiabatic inversion-recovery (STAIR) sequence. Magn Reson Med. 2022;88(3):1156-1169. doi:10.1002/mrm.29287

12. van der Weijden CWJ, García DV, Borra RJH, et al. Myelin quantification with MRI: A systematic review of accuracy and reproducibility. Neuroimage. 2021;226:117561. doi:10.1016/j.neuroimage.2020.117561

13. Mancini M, Karakuzu A, Cohen-Adad J, Cercignani M, Nichols TE, Stikov N. An interactive meta-analysis of MRI biomarkers of myelin. Jbabdi S, Baker CI, Jbabdi S, Does M, eds. eLife. 2020;9:e61523. doi:10.7554/eLife.61523

14. Lazari A, Lipp I. Can MRI measure myelin? Systematic review, qualitative assessment, and meta-analysis of studies validating microstructural imaging with myelin histology. Neuroimage. 2021;230:117744. doi:10.1016/j.neuroimage.2021.117744

15. Ilbey S, Antonia S, Uzay E, Michael B, Ali CÖ. Multi-Contrast Multi-Resolution UTE for Simultaneous Quantitative and High-Resolution MRI. In Proc Intl Soc Mag Reson Med 32. 5110. https://cds.ismrm.org/protected/23MPresentations/abstracts/5110.html.

16. Reisert M, Kellner E, Dhital B, Hennig J, Kiselev VG. Disentangling micro from mesostructure by diffusion MRI: A Bayesian approach. Neuroimage. 2017;147:964-975. doi:10.1016/j.neuroimage.2016.09.058

17. Peters AM, Brookes MJ, Hoogenraad FG, et al. T2* measurements in human brain at 1.5, 3 and 7 T. Magnetic Resonance Imaging. 2007;25(6):748-753. doi:10.1016/j.mri.2007.02.014

18. Ma D, Gulani V, Seiberlich N, et al. Magnetic resonance fingerprinting. Nature. 2013;495(7440):187-192. doi:10.1038/nature11971

Figures

Fig.1: Sequence diagram (a) For the first 2 echoes, a 1-1 binomial water-excited pulses (T_RF = 20µs, T_sat = 1.25 ms) and a dipolar gradient ( G_Echo,1 = -G_Echo,2= 20mT/m, T_Enc,1 = 900µs) are used. (b) A k-space trajectory will start and end at the center of k-space. For the third echo, G_Echo,3 and T_Enc,2 are determined based on resolution and bandwidth. (c) The high-resolution water-excited structural image was reconstructed by retrospectively merging the consecutive image sets before reconstruction. (d) The first 2 echoes of mcUTE with 9 different TEs to capture rapidly decaying signals.

Fig.2: Data post-processing workflow: images of different sequences were processed by FMRIB Software Library (FSL) including bias field correction, extraction of brain tissue, and registration to T1-weighted anatomy images. To obtain T2 information, a T2 SPACE sequence with varied TEs was utilized for myelin water fraction (MWF) imaging for evaluation. The signal intensities of the first two echoes of mcUTE with varied TEs were fitted to a three-component model to compute the us-T2* map and myelin fraction mapping, where the long T2* information was obtained from FLASH 3D.

Fig.3: In vivo measurements of one volunteer with mcUTE. (a) The first two echoes of mcUTE with different ultra-short TEs (TE1={20,40,60,80,100,140,180,220,260}µs, TE2=TE1+900µs, 1.71mm/1.71mm isotropic resolution) to capture rapidly decaying signals. (b) High-resolution (0.50mm isotropic resolution) water excitation structure images of the 3rd Echo of mcUTE.

Fig.4: Estimated myelin fractions and ultra-short T2* in different anatomical sections measuring in vivo on three healthy volunteers. All estimated results are shown as mean ± standard deviation.

Fig.5: Fitting results: (a) ultra-short T2* components map, (b) myelin fractions (MF) map, (c) long T2* value map, (d) myelin water fractions (MWF) map. Myelin fraction in white matter was higher than in gray matter, consistent with the myelin water fraction response. Long T2* map shows that GM has larger long T2* values compared to WM, which is consistent with previous reports.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)

4588

DOI: https://doi.org/10.58530/2024/4588