An approach to high resolution diffusion tensor imaging in fixed primate brain

Abstract

High resolution ex vivo diffusion tensor imaging (DTI) studies of neural tissues can improve our understanding of brain structure. In these studies we can modify the tissue relaxation properties of the fixed tissues to better suite the scanner hardware. We investigated the use of Gd-DTPA contrast agent to provide the optimum signal-to-noise (SNR) ratio in 3D DTI scans of formalin fixed nonhuman primate brains at 4.7 T. Relaxivity measurements in gray and white matter allowed us to optimize the Gd concentration for soaking the brains, resulting in a 2 fold improvement in SNR for the 3D scans. FA changed little with Gd concentrations up to 10 mM although ADC was reduced at 5 and 10 mM. Comparison of in vivo, fresh ex vivo and fixed brains showed no significant FA changes but reductions in ADC of about 50% in fresh ex vivo, and 64% and 80% in fixed gray and white matter respectively. Studies of the temperature dependence of diffusion in these tissues suggested that a 30° increase in sample temperature may yield an improvement of up to 55% in SNR-efficiency for a given diffusion weighting. Our Gd soaking regimen appeared to have no detrimental effect on standard histology of the fixed brain sections. Our methods yield both high SNR and spatial resolution DTI data in fixed primate brains, allowing us to perform high resolution tractography which will facilitate the process of ‘validation’ of DTI fiber tracts against traditional measures of brain fiber architecture.

Introduction

Numerous factors affect the final quality of diffusion tensor imaging (DTI) or high angular resolution diffusion imaging (HARDI) scans of brain tissue. Parameters such as spatial resolution, b-value, q-space sampling (e.g. spherical shells, Cartesian, etc.), pulse sequence type, brain coverage, etc. can be traded off against each other, however the ultimate constraints are the magnet’s gradient system and the total time available for the scan. The maximum gradient strength sets the relationship between the maximum b-value and TE, and hence links diffusion contrast with SNR. The gradient duty cycle limit constrains the number of slices (for multislice acquisitions) or the minimum TR (for 3D acquisitions). Typically therefore, one would scan using both the maximum gradient strength and maximum duty cycle available. As with all MRI scans, SNR scales with the total imaging time available. For fixed tissue samples this is limited only by the scanner's stability and local scheduling issues. As a result, there have been numerous high resolution structural MRI studies of fixed human and animal brains, both as a means of relating MRI contrast to standard pathology and histopathology (Augustinack et al., 2005, Boyko et al., 1994, Fatterpekar et al., 2002, Fernando et al., 2004, Jacobs et al., 1999, Johnson et al., 1993, Pfefferbaum et al., 2004) and as a method of morphological phenotyping in genetically modified animals (Benveniste and Blackband, 2002, Johnson et al., 2002a, Johnson et al., 2002b).

Most MR microscopy scans on fixed neural tissue derive image contrast from variations in T₂^* and proton density (Augustinack et al., 2005, Blamire et al., 1999, Fatterpekar et al., 2002), however there has been growing interest in the use of diffusion weighted imaging and DTI as an important source of contrast in fixed brains (D’Arceuil et al., 2005, de Crespigny et al., 2005b, Englund et al., 2004, Guilfoyle et al., 2003, Jacobs et al., 1999, Johnson et al., 1993, Kroenke et al., 2005, Kroenke et al., 2006, Larsson et al., 2004, Mori et al., 2001, Pfefferbaum et al., 2004, Schwartz et al., 2003, Shepherd et al., 2006, Shepherd et al., 2003a, Shepherd et al., 2003b, Sun et al., 2005, Sun et al., 2003, Tyszka et al., 2006, Verma et al., 2005, Yen et al., 2005). Evidence from rat brain (Sun et al., 2003), and human brain (Guilfoyle et al., 2003) and spinal cord (Mamata et al., 2006) suggests that while the apparent diffusion coefficient (ADC, taken hereafter to mean the mean diffusivity, or Trace/3 of the diffusion tensor), is reduced in fixed tissue, the fractional diffusion anisotropy (FA) is preserved, even after stroke (Sun et al., 2005). Another key advantage of ex vivo MRI of the brain is the ability to use exogenous contrast agents (e.g. Gd-DTPA) to enhance image contrast throughout the tissue (as opposed to just vessels and certain pathologies) (Benveniste and Blackband, 2002, Johnson et al., 1993). For DTI, where image contrast is not directly dependent upon tissue relaxation times, contrast agents can be used to permit more rapid scanning and/or improved SNR (D’Arceuil et al., 2005, Tyszka et al., 2006). Specifically, it allows us to optimize the tissue relaxation parameters to best suite the MRI scanner/imaging sequence, rather then only optimizing the pulse sequence to suit the sample as is the case in vivo.

Given the growing interest in diffusion MRI of fixed neural tissue, we have explored some of the major factors affecting image quality and SNR in such acquisitions, with the ultimate aim of generating optimum data for high resolution ex vivo diffusion tractography studies. In particular, we investigated the use of a Gd-based contrast agent for DTI in fixed primate brain, and compared in vivo and ex vivo DTI in the same brains. We also evaluated the effect of sample temperature upon the SNR of the DTI data, and studied the effect of the contrast agent on standard histopathological examination of brain sections.