Intravital 3D visualization and segmentation of murine neural networks at micron resolution

Abstract

Optical coherence tomography (OCT) allows label-free, micron-scale 3D imaging of biological tissues' fine structures with significant depth and large field-of-view. Here we introduce a novel OCT-based neuroimaging setting, accompanied by a feature segmentation algorithm, which enables rapid, accurate, and high-resolution in vivo imaging of 700 μm depth across the mouse cortex. Using a commercial OCT device, we demonstrate 3D reconstruction of microarchitectural elements through a cortical column. Our system is sensitive to structural and cellular changes at micron-scale resolution in vivo, such as those from injury or disease. Therefore, it can serve as a tool to visualize and quantify spatiotemporal brain elasticity patterns. This highly transformative and versatile platform allows accurate investigation of brain cellular architectural changes by quantifying features such as brain cell bodies' density, volume, and average distance to the nearest cell. Hence, it may assist in longitudinal studies of microstructural tissue alteration in aging, injury, or disease in a living rodent brain.

Figure 1

High-resolution cortical imaging by highly stable OCT system in vivo. (a) Schematic of our OCT imaging setup showing fixation of a mouse head by a customized head-clamp that is mounted on a translational (z-direction) stage. (b) A sampled 3D OCT volume (0.7 X 0.7 X 0.4 mm) of the mouse visual cortex (c) Sampled OCT b-scan, at the location marked by the dashed line in panel (b), of a cross-section of the five cortical layers (marked L1-L5) in the mouse visual cortex. Cell bodies appear as circular dark foci. (d–f) Sampled high-resolution en face images at cortical depth (d) 132 µm, (e) 255 µm, and (f) 380 µm [contrast in (d–f) was manually enhanced to emphasize cell bodies]. Red arrow points to a blood vessel, the white arrow points to CNS cell bodies, and the yellow arrow points to a myelinated axon.

Figure 2

Feature segmentation method allows for micron-scale 3D visualization of the cortex. (a) A sampled b-scan overlaid with the segmented CNS cell bodies; color encodes depth below the cranial window (b–d) Sampled en face images overlaid with the segmented CNS cell bodies at cortical depth (b) 132 µm, (c) 255 µm, and (d) 380 µm. (e) Visualization of a 3D CNS segmentation mask of brain neural networks in a living mouse at micron-level resolution, with a color-encoded depth of more than 10,000 CNS cell bodies, and their surrounding myelinated axons (grey); volume dimension is 0.7 X 0.7 X 0.4 mm (XYZ).

Figure 3

Quantifying cellular morphological features across the cortical column of mice in vivo. Four male mice were imaged by our OCT neuroimaging system and morphological traits were assessed and quantified: (a) average CNS cell body volume (b) average CNS cell body density (c) average distance to the nearest neighboring CNS cell body.

Figure 4

A cortical elasticity map assesses temporal cellular structure dynamics in vivo. (a) CNS cell bodies’ positional temporal changes, attributed to subtle tissue movement, between Day 1 and Day 10 of imaging as a function of the difference between rigid and elastic registration. Arrow’s color represents the spatial positional temporal changes, based on each individual cell body location change, while its length indicates the distance. (b–d) spatial positional temporal changes (distance) histogram of the center position offset of registered CNS cell bodies across X, Y, and Z axes between time points. The histogram shows only registered cells that were less than 10 µm apart, between (b) imaging Day 8 to Day 1 (76% of cells), (c) imaging Day 10 to Day 1 (76% of cells), and (d) imaging Day 10 to Day 8 (78% of cells).