High b‐value q‐space analyzed diffusion‐weighted MRI: Application to multiple sclerosis

Abstract

Multiple sclerosis (MS) is an inflammatory disease of the central nervous system (CNS) which affects nearly one million people worldwide, leading to a progressive decline of motor and sensory functions, and permanent disability. High b-value diffusion-weighted MR images (b of up to 14000 s/mm 2 ) were acquired from the brains of controls and MS patients. These diffusion MR images, in which signal decay is not monoexponential, were analyzed using the q-space approach that emphasizes the diffusion characteristics of the slow-diffusing component. From this analysis, displacement and probability maps were constructed. The computed q-space analyzed MR images that were compared with conventional T 1 , T 2 (fluid attenuated inversion recovery (FLAIR)), and diffusion tensor imaging (DTI) images were found to be sensitive to the pathophysiological state of white matter. The indices used to construct this qspace analyzed MR maps, provided a pronounced differentiation between normal tissue and tissues classified as MS plaques by the FLAIR images. More importantly, a pronounced differentiation was also observed between tissues classified by the FLAIR MR images as normal-appearing white matter (NAWM) in the MS brains, which are known to be abnormal, and the respective control tissues. The potential diagnostic capacity of high b-value diffusion q-space analyzed MR images is discussed, and experimental data that explains the consequences of using the q-space approach once the short pulse gradient approximation is violated are presented. Magn Reson Med 47:115-126, 2002.

Indirect measurement of regional axon diameter in excised mouse spinal cord with q-space imaging: Simulation and experimental studies

“…As a result, the QSI studies cited above either violated the SGPA or their displacement profile resolutions (ranging from 2 (Chin et al, 2004) to 20 (King et al, 1997) µm) exceeded typical axon diameters (1-2 µm). Mitra et al (Mitra and Halperin, 1995) predicted that as the gradient duration increased in relation to diffusion time, the displacement profile would artifactually narrow, which has been observed experimentally (Assaf et al, 2002a).…”

“…Other investigators have used the root-mean-squared (RMS) displacement instead of the displacement profile FWHM to estimate the mean axon diameter (Assaf et al, 2002a;Assaf et al, 2000;Biton et al, 2005). The RMS displacement is calculated easily from the displacement profile FWHM by multiplying it by a factor of 0.425 (Cory and Garroway, 1990).…”

“…In order to extend the 1D QSI technique to the tortuous WM tracts of the brain, Assaf et al (Assaf et al, 2002a) incorporated a tensor analysis of 1D QSI experiments performed along seven gradient directions to produce rotationally invariant parameters analogous to DTI. The tensor analysis allows for extraction of the displacement profile metrics corresponding to water diffusing perpendicular to the axon fiber tracts.…”

Reproducibility of quantitative structural and physiological MRI measurements

“…The PD model addresses a limitation of the standard DTI‐FA model, which assumes a single pool of anisotropically diffusing water. However, diffusion signal behaves as a biexponential function of b ‐values, representing two, unrestricted and restricted, “pools” of water ( M u and M r , respectively) (Assaf et al., 2002, 2005; Clark et al., 2002; Wu et al., 2011a,b). Parameters derived from the biexponential modeling, such as perfusion‐diffusivity index (PDI), are therefore sensitive to membrane permeability (Kochunov et al., 2013, 2014; Sukstanskii, Ackerman, & Yablonskiy, 2003; Sukstanskii, Yablonskiy, & Ackerman, 2004).…”

“…Parameters derived from the biexponential modeling, such as perfusion‐diffusivity index (PDI), are therefore sensitive to membrane permeability (Kochunov et al., 2013, 2014; Sukstanskii, Ackerman, & Yablonskiy, 2003; Sukstanskii, Yablonskiy, & Ackerman, 2004). In short, diffusion images were preprocessed to perform an region of interest (ROI) based fit for a two‐compartment diffusion model (Equation (1)) that assumed that intravoxel signal is formed by a contribution from two compartments (Assaf et al., 2002, 2005; Clark et al., 2002; Panagiotaki et al., 2009; Wu et al., 2011a,b). $Sfalse(bfalse)=S_0·false(M_{\mathrm{u}}·{\mathrm{e}}^{-b{D}_u}+(1-M_{\mathrm{normalu}})·{\mathrm{e}}^{-b{D}_r}false)$ $\text{PDI}=\frac{D_{\mathrm{r}}}{D_{\mathrm{u}}}$where S ( b ) is the average diffusion‐weighted signal for a given b ‐value, averaged across all directions.…”