Considerations and recommendations from the ISMRM Diffusion Study Group for preclinical diffusion MRI: Part 3-Ex vivo imaging: Data processing, comparisons with microscopy, and tractography

Abstract

Preclinical diffusion MRI (dMRI) has proven value in methods development and validation, characterizing the biological basis of diffusion phenomena, and comparative anatomy. While dMRI enables in vivo non-invasive characterization of tissue, ex vivo dMRI is increasingly being used to probe tissue microstructure and brain connectivity. Ex vivo dMRI has several experimental advantages that facilitate high spatial resolution and high SNR images, cutting-edge diffusion contrasts, and direct comparison with histological data as a methodological validation. However, there are a number of considerations that must be made when performing ex vivo experiments. The steps from tissue preparation, image acquisition and processing, and interpretation of results are complex, with many decisions that not only differ dramatically from in vivo imaging of small animals, but ultimately affect what questions can be answered using the data. This work concludes a three-part series of recommendations and considerations for preclinical dMRI. Herein, we describe best practices for dMRI of ex vivo tissue, with a focus on image pre-processing, data processing, and comparisons with microscopy. In each section, we attempt to provide guidelines and recommendations but also highlight areas for which no guidelines exist (and why), and where future work should lie. We end by providing guidelines on code sharing and data sharing and point toward open-source software and databases specific to small animal and ex vivo imaging.

Keywords: acquisition; best practices; diffusion MRI; diffusion tensor; ex vivo; microstructure; open science; preclinical; processing; tractography.

FIGURE 1

There are many artifacts that must be corrected for in preprocessing. These are not necessarily presented in order, and correction may not be necessary in all cases. Nevertheless, the most common order for pre‐processing steps (after data important and quality check) is: (i) thermal noise reduction (referred to as denoising), (ii) Gibbs ringing correction, (iii) susceptibility distortion + motion + eddy current corrections (+ gradient non‐linearity, if applicable), (iv) Rician bias correction and (v) signal drift correction. Figures kindly provided by Ileana Jelescu, Kurt Schilling, or reproduced from.,

FIGURE 2

Ex vivo maps from DTI, DKI, and a biophysical model (white matter tract imaging; WMTI). Parameter maps show radial diffusivity (DTI), radial kurtosis (DKI), and radial extra‐axonal diffusivity (WMTI) for control mice and two hypomyelinated mouse models (Rictor and TSC). Ex vivo imaging was performed on a 15.2T Bruker Biospec scanner at 150 μm isotropic resolution using a 3D diffusion‐weighted fast spin‐echo and b‐values of 3000 and 6000 s/mm². Figure reproduced from Ref. .

FIGURE 3

Ex vivo tractography on mouse (left), macaque (middle), and human brains (right). In the mouse, high resolution tractography was used to identify region‐to‐region differences in connectivity in models of Huntington's disease; here, tractography is able to delineate Striatal connectivity. In the Macaque brain, standardized protocols were developed to enable robust and automated segmentation of 42 white matter pathways. In the human brain, diffusion data at high spatial resolution showed feasibility of reconstructing brainstem nuclei and white matter of the brainstem.

FIGURE 4

Examples of co‐localized MRI and histological data. Top: Fixed multiple sclerosis human spinal cord; bottom: Fixed mouse liver. From left to right: B = 0 image; whole‐sample histological section taken within the tissue corresponding to the MRI slice (proteolipid protein [PLP] immunostain for the spinal cord; hematoxylin and eosin [HE] staining for the mouse liver); dMRI parametric map (fractional anisotropy [FA] for the spinal cord; ADC for the mouse liver); histological parametric maps co‐registered to dMRI space (myelin staining fraction for the spinal cord, and volume‐weighted cell size statistics for the mouse liver, evaluated within histological image patches matching the in‐plane MRI resolution). The data reproduced in this figure with kind permission from C.A.M. Gandini Wheeler‐Kingshott, G.C. DeLuca and R. Perez‐Lopez refer to previous dMRI studies.,

FIGURE 5

Histology to diffusion MRI alignment example using the intermediate modality, block‐face images. The 2D histology (A1) can be mapped to specific 2D block‐face images (A2), which can be stacked into a 3D volume and mapped directly to 3D diffusion MRI data/derived data (A3). In this example, fiber orientation distribution from histology is aligned with similar measures estimated from diffusion MRI for validation purposes., Insets (A4–A5) zoom in to show alignment across modalities, with final panels showing histology‐derived (left) and MRI derived (right) fiber orientation distributions.

FIGURE 6

Examples of direct imaging of histological slices. (Left) A Black‐Gold II stained histology (myelin‐stain) was directly imaged in a microsurface coil with in‐plane resolution of 15.6um. Tractography and red‐green‐blue orientation maps are shown overlaid on myelin stain. (Top right) a 25 μm‐thick Nissl‐stained rat spinal cord tissue was imaged using a microsurface coil at 7.8 μm in‐plane resolution with corresponding diffusion weighted image showing excellent correspondence. (Bottom right) A 300 μm‐thick rat hippocampal slice is imaged under a light‐microscope and imaged using a slide‐mounted microsurface coil at 12.5 μm in‐plane resolution to study the microstructural effects of diffusion times and b‐values. Images reproduced from (left), (top right), (bottom right).

FIGURE 7

Left: Tractography validation via anatomical tracers (top, ¹²² ) and microdissection (bottom, ¹²³ ). Right: Recent advances in high throughput 3D imaging of anatomical tracers facilitates the tracking of single neurons, here projecting from the medial dorsal nucleus of the thalamus.

FIGURE 8

Fiber orientations extracted from various microscopy modalities can be used to validate biophysical models or fiber reconstruction methods in diffusion MRI. These modalities include structure tensor analysis of histological sections (myelin ST ¹⁰⁶ ) and polarized light imaging (PLI ¹⁰⁶ ), small angle x‐ray scattering (SAXS ¹²⁵ ) and scattered light imaging (SLI ¹²⁵ ), micro‐CT (u‐CT ⁸⁷ ), and EM.