Perivascular network segmentations derived from high-field MRI and their implications for perivascular and parenchymal mass transport in the rat brain

Abstract

A custom segmentation workflow was applied to ex vivo high-field MR images of rat brains acquired following in vivo intraventricular contrast agent infusion to generate maps of the perivascular spaces (PVS). The resulting perivascular network segmentations enabled analysis of perivascular connections to the ventricles, parenchymal solute clearance, and dispersive solute transport within PVS. Numerous perivascular connections between the brain surface and the ventricles suggest the ventricles integrate into a PVS-mediated clearance system and raise the possibility of cerebrospinal fluid (CSF) return from the subarachnoid space to the ventricles via PVS. Assuming rapid solute exchange between the PVS and CSF spaces primarily by advection, the extensive perivascular network decreased the mean clearance distance from parenchyma to the nearest CSF compartment resulting in an over 21-fold reduction in the estimated diffusive clearance time scale, irrespective of solute diffusivity. This corresponds to an estimated diffusive clearance time scale under 10 min for amyloid-beta which suggests that the widespread distribution of PVS may render diffusion an effective parenchymal clearance mechanism. Additional analysis of oscillatory solute dispersion within PVS indicates that advection rather than dispersion is likely the primary transport mechanism for dissolved compounds greater than 66 kDa in the long (> 2 mm) perivascular segments identified here, although dispersion may be significant for smaller compounds in shorter perivascular segments.

Figure 1

Partial maximum intensity projections (pMIPs). Each panel is a projection across 30 slices parallel to the coronal (a–h), sagittal (i–n), and transverse (o–t) planes for Rat 3 (a–d, i–k, and o–q) and Rat 4 (e–h, l-n, and r–t). Proceeding alphabetically, each coronal panel is posterior to the preceding panel, each sagittal panel is lateral to the preceding panel, and each transverse panel is ventral to the preceding panel for each animal. Notable vessels are labeled (see Supplementary Table S1).

Figure 2

Perivascular and ventricle segmentations. The PVS (gold) and the ventricles (blue) segmented in two sets of 30 contiguous coronal slices are overlayed on the corresponding pMIPs for Rat 3 (a, c) and Rat 4 (h, j). These pMIPs are shown with overlayed segmentations for comparison with Fig. 1b, c, f, g, respectively. These 30-slice segmentations are presented alongside their 3D renderings for Rat 3 (b, d) and Rat 4 (i, k). All PVS (gold) and the ventricles (blue) are rendered in 3D for Rat 3 (e–g) and Rat 4 (l–n). Panels (e) and (l) are a rostral view of the coronal plane, (f, m) are a dorsal view of the transverse plane, and (g) and (n) are a left view of the sagittal plane. Notable vessels are labeled (see Supplementary Table S1).

Figure 3

Connected perivascular spaces occupying periventricular regions. The portions of the perivascular network containing segments that traverse a three-voxel margin around the ventricles are 3D-rendered for Rat 3 (a–c) and Rat 4 (d–f). The left column is a rostral view of the coronal plane, the central column is a dorsal view of the transverse plane, and the right column is a left view of the sagittal plane. The needle track in each rat is labeled with a red star. Notable vessels are labeled (see Supplementary Table S1).

Figure 4

Minimum distance between parenchyma and CSF spaces. The distances between parenchyma voxels and the nearest CSF space without PVS (a–c) and including PVS (d–f) for Rat 4 are presented as planar color images. The left column is a rostral view of the coronal plane, the central column is a dorsal view of the transverse plane, and the right column is a left view of the sagittal plane. (g) Distributions of minimum distance between parenchyma voxels and the CSF spaces with and without PVS. (h) Diffusive transport time scales for a range of solute effective diffusivities with and without PVS.

Figure 5

Oscillatory solute dispersion in PVS. (a) Dispersion model geometry consisting of a well-mixed CSF region with steady concentration $C_1$ and an adjacent semi-infinite straight perivascular segment initially at concentration $C$ = 0. (b) Measured lengths for a sample (n = 10) of the longest visually identifiable perivascular segments in Rat 5. The data points have been horizontally scattered to improve visibility. (c) Position and (d) velocity of advancing tracer fronts with concentrations $\alpha C_1$ over time (k = 1.05). (e) Time to traverse 250 μm and 1000 μm for a range of solute molecular diffusivities. Solid and dashed lines denote a 5% (k = 1.05) and 70% (k = 1.7) increase in the diffusion coefficient (dispersive enhancement) caused by oscillatory flow.

Figure 6

Image analysis workflow. Following the 24 h MRI session at 17.6 T, the brain was extracted and visualized with a custom maximum intensity projection technique (blue box). Then, brain regions and the PVS were segmented and rendered in 3D (red box). Lastly, PVS-ventricle connectivity and transport were analyzed (green box). The software applications used in each step, as indicated by the superscript letters, are as follows: (a) rodent Brain Extraction Tool (rBET); (b) FMRIB's Linear Registration Tool (FLIRT); (c) MATLAB r2018a; (d) ImageJ; (e) itk-SNAP.

Figure 7

Signal intensity and tubeness distributions. (a) Signal intensity from the brains with contrast agent (Rat 1–5) and without contrast agent (Naïve 1–2). The signal intensity for each animal was mapped linearly between 0 and 1. (b) Signal intensity following intensity range remapping to align signal distribution peaks and create better perivascular contrast with the surrounding tissue. (c–d) Tubeness distribution for the brain (c) interior and (d) surface. The gray vertical line is the average of the tubeness values corresponding to the 0.95 cumulative distribution value in each naïve brain. This tubeness value is greater than the tubeness in 95% of the naïve voxels on average.

Figure 8

Brain volumes of interest (VOI) for perivascular space segmentation analysis. A coronal slice from the VOI in the (a) cortex, (b) striatum, (c) periaqueductal tissue, and (d) brain stem. Each VOI spans 30 coronal slices.