5036

The role of perivascular spaces in white matter dynamic susceptibility contrast MRI
Jonathan Doucette1,2, Enedino Hernández-Torres1,3, Christian Kames1,2, and Alexander Rauscher1,3,4

1UBC MRI Research Centre, Vancouver, BC, Canada, 2Physics and Astronomy, University of British Columbia, Vancouver, BC, Canada, 3Pediatrics, University of British Columbia, Vancouver, BC, Canada, 4Division of Neurology, Faculty of Medicine, University of British Columbia, Vancouver, BC, Canada

Synopsis

Using vascular parameters obtained from dynamic susceptibility contrast MRI, the gradient echo (GRE) and spin echo (SE) dynamic susceptibility contrast (DSC) induced changes in $$$\Delta{R_2^{(*)}}$$$ were simulated at 3T in order to investigate the effects of tissue orientation and perivascular spaces (PVS). We found that the orientation dependence of both $$$\Delta{R_2}$$$ and $$$\Delta{R_2^*}$$$ are amplified by PVS, though $$$\Delta{R_2}$$$ is far more sensitive to PVS.

Introduction

Perfusion in the cerebral white matter (WM) as measured by both spin echo (SE) and gradient echo (GRE) dynamic susceptibility contrast (DSC) has been demonstrated to be highly dependent on the angle $$$\alpha$$$ between the WM tract and the main magnetic field $$$B_0$$$1,2. This is due to the spatially anisotropic vascular structure of the WM, which is known to contain large vessels which run in parallel with the WM fibre tracts3. Moreover, perivascular spaces (PVS) surround large blood vessels throughout the brain. These PVS are two to three times the diameter of the containing vessel, and are often enlarged in pathological conditions and aging4,5. Recent work demonstrated that free water in the PVS significantly affects DTI measures6, suggesting that this fluid cannot be neglected in DSC measurements due to the influence that diffusion mediated dephasing has been shown to have on DSC measurements, particularly in SE DSC2,7,8. Here, we perform numerical simulations of the GRE DSC and SE DSC signals in order to explore the influence of PVS on the tissue orientation dependent measurements.

Methods

DSC data from 10 healthy subjects was acquired at 3T using SE EPI (TR/TE=1530/60ms, voxel volume=3x3x3mm3)2, and from 13 subjects with multiple sclerosis using GRE EPI (TR/TE=2417ms/40ms, voxel volume=1.5x1.5x4mm3)1. Fibre angle was calculated using DTI scans as described in9. The DSC signal in each WM voxel for each data set was sorted according to fibre angle $$$\alpha$$$ into bins of width $$$5^\circ$$$ ranging from $$$2.5^\circ$$$ to $$$87.5^\circ$$$. $$$\Delta{R_2^{(*)}}=-\frac{1}{\text{TE}}\log\left(\frac{S}{S_0}\right)$$$ was calculated and averaged across all voxels in each bin for all subjects. $$$S_0$$$ and $$$S$$$ are the baseline signal and signal corresponding to peak contrast agent, respectively. The SE DSC and GRE DSC experiments were simulated for echo times of 60ms and 40ms, respectively.

Numerical simulation of the SE and GRE DSC signals was performed by solving the Bloch-Torrey equation within simulated WM voxels filled with anisotropic and isotropic blood vessels, as described in our previous work2. The simulated vascular architecture shown in Figure 1(a) resulted from a parameter fitting process2, matching simulated $$$\Delta{R_2}$$$ vs. WM fibre angle $$$\alpha$$$ curves to measured $$$\Delta{R_2}$$$ data. Figure 1(b) shows the corresponding induced field inhomogeneities, calculated by convolution of the susceptibility distribution with a magnetic unit dipole. A similar geometry and inhomogeneity map was obtained from the $$$\Delta{R_2^*}\;\text{vs.}\;\alpha$$$ data. In both geometries, cylindrical PVS were placed around the anisotropic vessels with fixed radii of $$$\rho=2X$$$ relative to the containing vessels. This corresponds to PVS volume of $$$\nu=3X$$$ relative to the anisotropic blood volume, where $$$\nu=\frac{\pi(\rho{R})^2{L}-\pi{R}^2{L}}{\pi{R}^2{L}}=\rho^2-1$$$, with $$$R,L$$$ the anisotropic vessel radius and length.

These simulated geometries were used as inputs for the forward calculation of the $$$\Delta{R_2^{(*)}}\;\text{vs.}\;\alpha$$$ curves, wherein $$$\nu$$$ was varied from $$$0X$$$ to $$$10X$$$. The diffusivity within the simulated voxel was set to that of water, $$$3037\mu{m}^2/ms$$$; while this value is too high for WM tissue, it nevertheless results in a good approximation for the $$$\Delta{R_2^*}\;\text{vs.}\;\alpha$$$ curves, as orientation-dependent diffusive dephasing occurs in the immediate vicinity of the anisotropic vessels, i.e. within the PVS, wherein diffusivity is high2.

Results

Increased PVS relative volume resulted in increased simulated $$$\Delta{R_2^{(*)}}$$$ curves as functions of WM fibre angle $$$\alpha$$$. In both cases, greater increase was found for larger $$$\alpha$$$. The SE DSC signal was found to be more sensitive to small PVS volumes than was GRE DSC. For large relative volumes ($$$\nu>3X$$$) SE DSC becomes relatively insensitive to PVS volume changes, whereas the GRE DSC signal becomes more sensitive.

Discussion and Conclusion

While increased PVS volume increased both $$$\Delta{R_2}$$$ and $$$\Delta{R_2^*}$$$ curves as functions of WM fibre angle, the mechanisms providing the change are manifestly different for SE DSC compared to GRE DSC. In SE DSC, larger PVS – which have much longer $$$T_2$$$ than the surrounding tissue – create larger regions of longer lasting signal surrounding the large anisotropic vessels. In these regions, the diffusive spin dephasing within the local anisotropic field inhomogeneities, which drive $$$\Delta{R_2}$$$, is increased. Naturally, this effect increases with increasing anisotropy of the local fields, which increases with $$$\alpha$$$.

For GRE DSC, $$$\Delta{R_2^*}$$$ is driven primarily by static dephasing, with diffusion providing a smaller secondary mechanism8. For this reason, the additional diffusive dephasing provided by increased PVS volume has little effect for small PVS volumes (below $$$\nu\approx{3X}$$$). Above this threshold, however, $$$\Delta{R_2^*}$$$ begins to increase due to increased long lasting signal in the immediate vicinity of the anisotropic vessels. Note that PVS relative radii of $$$2X$$$($$$3X$$$) correspond to relative volumes of $$$3X$$$($$$8X$$$) and that local field inhomogeneities decay as $$$1/r^2$$$ away from the cylinder axis, and so considerable anisotropic field moves inside the PVS when $$$\rho$$$ increases from $$$2X$$$ to $$$3X$$$. This results in $$$\alpha$$$-dependent increases in $$$\Delta{R_2^*}$$$.

Acknowledgements

No acknowledgement found.

References

  1. Hernández-Torres, E., Kassner, N., Forkert, N.D., Wei, L., Wiggermann, V., Daemen, M., Machan, L., Traboulsee, A., Li, D., Rauscher, A., 2017. Anisotropic cerebral vascular architecture causes orientation dependency in cerebral blood flow and volume measured with dynamic susceptibility contrast magnetic resonance imaging. J. Cereb. Blood Flow Metab. Off. J. Int. Soc. Cereb. Blood Flow Metab. 37, 1108–1119. https://doi.org/10.1177/0271678X16653134
  2. Doucette, J., Wei, L., Hernández-Torres, E., Kames, C., Forkert, N.D., Aamand, R., Lund, T.E., Hansen, B., Rauscher, A., 2018. Rapid solution of the Bloch-Torrey equation in anisotropic tissue: Application to dynamic susceptibility contrast MRI of cerebral white matter. NeuroImage. https://doi.org/10.1016/j.neuroimage.2018.10.035
  3. Nonaka, H., Akima, M., Hatori, T., Nagayama, T., Zhang, Z., Ihara, F., 2003. Microvasculature of the human cerebral white matter: Arteries of the deep white matter. Neuropathology 23, 111–118. https://doi.org/10.1046/j.1440-1789.2003.00486.x
  4. Doubal, F.N., MacLullich, A.M.J., Ferguson, K.J., Dennis, M.S., Wardlaw, J.M., 2010. Enlarged Perivascular Spaces on MRI Are a Feature of Cerebral Small Vessel Disease. Stroke 41, 450–454. https://doi.org/10.1161/STROKEAHA.109.564914
  5. Mestre, H., Kostrikov, S., Mehta, R.I., Nedergaard, M., 2017. Perivascular spaces, glymphatic dysfunction, and small vessel disease. Clin. Sci. 131, 2257–2274. https://doi.org/10.1042/CS20160381
  6. Sepehrband, F., Cabeen, R.P., Choupan, J., Barisano, G., Law, M., Toga, A.W., Initiative, the A.D.N., 2018. A systematic bias in DTI findings. bioRxiv 395012. https://doi.org/10.1101/395012
  7. Doucette, J., Wei, L., Kames, C., Hernández-Torres, E., Aamand, R., Lund, T.E., Hansen, B., Rauscher, A. Anisotropic cerebral vascular architecture causes orientation dependency in cerebral blood flow and volume measured with spin echo dynamic susceptibility contrast magnetic resonance imaging, in: Proceedings of the 25th Annual Meeting of the International Society for Magnetic Resonance in Medicine (ISMRM), Honolulu, Hawaii, USA, April 2017, Oral Presentation.
  8. Doucette, J., Hernández-Torres, E., Kames, C., Wei, L., Rauscher, A. The effects of diffusion on the tissue orientation dependent gradient echo dynamic susceptibility contrast signal, in: Proceedings of the 26th Annual Meeting of the International Society for Magnetic Resonance in Medicine (ISMRM), Paris, France, June 2018, Oral Presentation.
  9. Hernández-Torres, E., Wiggermann, V., Hametner, S., Baumeister, T.R., Sadovnick, A.D., Zhao, Y., Machan, L., Li, D.K.B., Traboulsee, A., Rauscher, A., 2015. Orientation Dependent MR Signal Decay Differentiates between People with MS, Their Asymptomatic Siblings and Unrelated Healthy Controls. PLOS ONE 10, e0140956. https://doi.org/10.1371/journal.pone.0140956

Figures

Visualization of the simulation domain. (a) An example voxel geometry with three large anisotropic vessels ($$$R=89.9\mu{m}$$$) amid a bed of smaller spatially isotropic vasculature ($$$R=7.0\mu{m}\pm0.5\mu{m}$$$), and (b) a cross section of the corresponding local field inhomogeneities for $$$B_0$$$ perpendicular to the anisotropic vessels. The addition of a contrast agent at $$$4.00mM$$$ has increased the induced resonance frequencies from approximate peak values of $$$\pm150\text{rad}/s$$$ to $$$\pm600\text{rad}/s$$$.

Simulated $$$\Delta{R_2}$$$ vs. WM fibre angle $$$\alpha$$$ curves for perivascular space relative volumes $$$\nu$$$ ranging from $$$0X$$$ to $$$10X$$$. Increasing PVS relative volumes increases $$$\Delta{R_2}$$$ due to longer lived diffusion mediated dephasing in the strong field gradients nearest the anisotropic vessels, until a critical point ($$$\approx{3X}$$$) where the anisotropic fields have decayed sufficiently. Note that >97% of data points have an angle with $$$B_0$$$ greater than $$$15^\circ$$$; below $$$15^\circ$$$ there is an upward trend, which may be an artifact of the small number of voxels, which the model could not describe; a similar discrepancy is shown in Figure 3.

Simulated $$$\Delta{R_2^*}$$$ vs. WM fibre angle $$$\alpha$$$ curves for perivascular space relative volumes $$$\nu$$$ ranging from $$$0X$$$ to $$$10X$$$. The addition of PVS has relatively little affect on $$$\Delta{R_2^*}$$$ for small $$$\nu$$$ due to the GRE DSC signal being relatively insensitive to diffusion mediated dephasing. Eventually, for large $$$\nu>3X$$$, the amount of anisotropic field inhomogeneities in the PVS becomes large enough to induce significant additional signal loss, and $$$\Delta{R_2^*}$$$ increases. Note that, as with SE DSC simulations, data points below $$$15^\circ$$$ were not explainable by the model, likely due to the small number of voxels contributing to those angle bins.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
5036