Synopsis
Characterizing white matter (WM) microstructure in terms of axon size and volume fraction remains elusive and typically requires extremely strong gradients to resolve water diffusion in micron-sized axons. Here, we develop and apply a new approach for quantifying WM microstructure using a simple multi-gradient-echo (MGE) sequence. We use rat spinal cord as our experimental system, and show that, at long echo times, oscillations in the signal decay emerge, which can be used to robustly quantify the WM microstructure. The ensuing parameteric maps segment the SC into its underlying microstructures. The simplicity of the approach bodes well for many future applications.Purpose
Characterizing white matter (WM) microstructure noninvasively is paramount since action potential propagation velocities depend on axonal sizes (1,2), and since aberrations in axonal microstructures occur in neurodegenerative diseases (5). So far, such characterizations typically relied on
q-space imaging (QSI) (4,9,10) requiring very strong gradients, or oscillating gradient waveforms (14). Here, we develop a simple, multi-gradient-echo (MGE) methodology at ultrahigh field, which–by virtue of signal oscillations observed for the first time–is capable of reflecting WM microstructure.
Methods
All experiments were performed using an ultrahigh-field 16.4T Bruker Aeon Ascend magnet (700MHz proton frequency) equipped with gradients capable of producing up to 3000mT/m in every direction. Long-Evans rats, 14-16 week old, were perfused with 4% PFA and the cervical spinal cord (SC) was isolated (n=8), washed in PBS, and aligned with the main magnetic field. T2* weighted images were acquired using a MGE sequence with 60 echoes (TR = 200 ms, TE1 = 1.4 ms, echo spacing = 1.1ms, flip angle = 26°). Images were acquired with a resolution of 34x34x250 µm3 (380 averages, ~2h). The images were fit voxel wise to a simple two-component model taking into account an intra-axonal fraction on-resonance and another fraction representing the rest of spins (e.g., extra-cellular, myelin, etc.) fb=1-fa with a frequency offset Δω, arising from a presumed susceptibility difference Δχ=χext-χint. The signal was therefore modeled as:
$$\frac{S(TE)}{S_{0}}=\parallel f_{a}\cdot e^{-\frac{TE}{T_{2a}}}+(1-f_{a})\cdot e^{-TE\cdot (\frac{1}{T_{2b}}+i\Delta\omega)}\parallel$$
where T2i is the respective T2 of the ith component. Since the axons are nearly perfectly parallel to the main magnetic field (as validated via DTI, which found >90% of white matter voxels to be aligned within ±5 degrees of B0), we can neglect the axon-size dependent terms (6) that would otherwise be introduced into the Δω term in Eq 1. Magnitude TE-dependent MRI images were regressed onto Equation 1 via a non-negative least squares function in Matlab®. To corroborate our results, immunohistochemstry (IHC) targeting myelin (myelin proteolipid protein), intra-axonal space (neurofilament (NF) 160/200), and cell-bodies (DAPI) was performed using 50 µm free floating slices taken from the approximate imaged locations. The ensuing IHC images were then quantified for regional size distributions and axon packing.
Results
As predicted by Eq. 1, we find pronounced oscillating signals in all seven microstructurally distinct areas of the SC WM (Fig. 1). These signals were highly reproducible across all SCs, and different regions exhibited clearly distinct oscillations (Figure 1B, inset). The fit to equation 1 was excellent (Figure 1B, solid lines). Parameter maps derived from the voxel wise fit revealed dramatic contrasts in the SC (Figure 2): most notably, the largest frequency shift occurs in the dCST, the region with smallest axons, whereas the smallest frequency shifts occur in the VST – the area comprising the largest axons (Fig. 2A). The intra-axonal volume fractions (Fig. 2B) are in good agreement with the intra-axonal fraction as extracted from IHC, (Fig. 3). The relaxation parameters T
2a and T
2b, corresponding to the intra and all other spaces, respectively, seem to show less selectivity towards the microstructurally distinct regions. Even when data with much lower spatial resolution was acquired, these oscillations persisted (not shown).
Discussion
We have here demonstrated the emergence of oscillations in TE-dependent T
2*-weighted images, and showed that they can be used to resolve regional SC micro-architectures (Figure 2). In principle, oscillating signals have great value as they lend themselves towards spectral Fourier-based analyses rather than ill-posed Laplace inversions. The very existence of oscillations in these signals is valuable also in suggesting that many spins are effectively in the static dephasing regime and less so in the diffusion narrowing regime. In our case, IHC revealed that f and Δω reflect the regional axonal diameter and density, which were furthermore in good agreement with previously published work (9,10,14). Simulations show that these oscillations will persist in other angles other than zero, especially in real-valued data, making this approach appealing for a general characterization of the white matter morphology (not shown) and elucidation of the origins of these T
2* heterogeneities (3,7,8,11-13,15,16). The simplicity and rapid acquisition of MGE sequences, and the relatively weak gradients here used to deliver information about microstructure, are very appealing towards clinical applications of the methodology.
Conclusion
We presented a new simple methodology reporting on WM microstructure based on MGE sequences at ultrahigh fields. We showed a highly reproducible signal oscillation that can be fitted to a simple model to derive parametric maps, which strongly correlate with the regional axonal diameter and packing density. IHC corroborated our noninvasive findings.
Acknowledgements
The study was supported by funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 657366. SJ acknowledges Lundbeck Foundation grant R83-A7548.References
1. Caminiti R, Carducci F, Piervincenzi C, Battaglia-Mayer A, Confalone G, Visco-Comandini F, Pantano P, and Innocenti GM. Diameter, Length, Speed, and Conduction Delay of Callosal Axons in Macaque Monkeys and Humans: Comparing Data from Histology and Magnetic Resonance Imaging Diffusion Tractography. Journal of Neuroscience 2013;33:14501-14511.
2. Caminiti R, Ghaziri H, Galuske R, Hof PR, and Innocenti GM. Evolution amplified processing with temporally dispersed slow neuronal connectivity in primates. Proceedings of the National Academy of Sciences of the United States of America 2009;106:19551-19556.
3. Chen WC, Foxley S, and Miller KL. Detecting microstructural properties of white matter based on compartmentalization of magnetic susceptibility. Neuroimage 2013;70:1-9.
4. Cohen Y and Assaf Y. High b-value q-space analyzed diffusion-weighted MRS and MRI in neuronal tissues - a technical review. NMR in Biomedicine 2002;15:516-542.
5. Evangelou N, Konz D, Esiri MM, Smith S, Palace J, and Matthews PM. Size-selective neuronal changes in the anterior optic pathways suggest a differential susceptibility to injury in multiple sclerosis. Brain 2001;124:1813-1820.
6. Haacke EM, Brown RW, Thompson MR, and Venkatesan R. Magnetic Resonance Imaging: Physical Principles and Sequence Design. 1999.
7. Li W, Wu B, Avram AV, and Liu CL. Magnetic susceptibility anisotropy of human brain in vivo and its molecular underpinnings. Neuroimage 2012;59:2088-2097.
8. Luo J, He X, and Yablonskiy DA. Magnetic Susceptibility Induced White Matter MR Signal Frequency Shifts-Experimental Comparison Between Lorentzian Sphere and Generalized Lorentzian Approaches. Magnetic Resonance in Medicine 2014;71:1251-1263.
9. Ong HH and Wehrli FW. Quantifying axon diameter and intra-cellular volume fraction in excised mouse spinal cord with q-space imaging. Neuroimage 2010;51:1360-1366.
10. Ong HH, Wright AC, Wehrli SL, Souza A, Schwartz ED, Hwang SN, and Wehrli FW. Indirect measurement of regional axon diameter in excised mouse spinal cord with q-space imaging: Simulation and experimental studies. Neuroimage 2008;40:1619-1632.
11. Sati P, Silva AC, van Gelderen P, Gaitan MI, Wohler JE, Jacobson S, Duyn JH, and Reich DS. In vivo quantification of T-2* anisotropy in white matter fibers in marmoset monkeys. Neuroimage 2012;59:979-985.
12. Sati P, van Gelderen P, Silva AC, Reich DS, Merkle H, de Zwart JA, and Duyn JH. Micro-compartment specific T*(2) relaxation in the brain. Neuroimage 2013;77:268-278.
13. van Gelderen P, de Zwart JA, Lee J, Sati P, Reich DS, and Duyn JH. Nonexponential T2*decay in white matter. Magnetic Resonance in Medicine 2012;67:110-117.
14. Xu JZ, Li H, Harkins KD, Jiang XY, Xie JP, Kang H, Does MD, and Gore JC. Mapping mean axon diameter and axonal volume fraction by MRI using temporal diffusion spectroscopy. Neuroimage 2014;103:10-19.
15. Yablonskiy DA, Luo J, Sukstanskii AL, Iyer A, and Cross AH. Biophysical mechanisms of MRI signal frequency contrast in multiple sclerosis. Proceedings of the National Academy of Sciences of the United States of America 2012;109:14212-14217.
16. Yablonskiy DA, Luo J, Sukstanskii AL, Iyer A, and Cross AH. Biophysical mechanisms of MRI signal frequency contrast in multiple sclerosis. Proceedings of the National Academy of Sciences of the United States of America 2012;109:14212-14217.