Diffusion-weighted imaging is the only technique available today to probe the microscopic structural organization of the brain tissue, in vivo. Consequently, we will define a high field imaging protocol providing high angular and spatial resolution diffusion-weighted data that will be used for two complementary purposes:
- Estimation at each point of the white matter a local distribution of the orientations in order to map the anatomical connectivity of the SN, the STN, the RN and the PPN.
- Estimation at each point of the nuclei involved in Parkinsonian syndromes (SN, STN, RN, PPN) the characteristics of the diffusion process in order to infer some fine parcelations of these nuclei stemming from their cytoarchitectony.
This high field protocol will rely on a large number of diffusion-sensitization orientations (presumably 64 uniformly distributed divided in two sets of 32 complementary orientations) and will rely on the use of a large b-value (between 2000 s/mm2 and 4000 s/mm2) that is known to yield high angular resolution.
Any disorder of these anatomical entities, like abnormal morphometry or abnormal MRI contrast, represents a putative biomarker of the pathology.
The first step of analysis of the DWI data that remains close to the image acquisition corresponds to the construction of local robust models of the orientation distribution of the underlying structures. At the current resolution of DW-MRI, research groups agree that there are between one and two thirds of imaging voxels in the human brain white matter that contain fibre crossing bundles (Behrens et al., 2007). We know that in these locations, the diffusion is non-Gaussian and the diffusion tensor (DT) (Basser et al., 1994) is limited due to its intrinsic Gaussian diffusion assumption. Hence, DT-based tractography algorithms can follow false tracts and produce unreliable tracking results. To overcome limitations of the DT, several high angular resolution diffusion imaging (HARDI) techniques (Tuch et al., 2002; Alexander, 2005) have been proposed to estimate the diffusion orientation distribution function (ODF) (Tuch, 2004) of water molecules or other high order spherical function estimate of the diffusion profile (Jansons et al., 2003; Tournier et al., 2004; Alexander, 2005; Anderson, 2005, Ozarslan et al., 2006; Sakaie et al., 2007; Tournier et al., 2007, Dellacqua et al., 2007; Kaden et al., 2007; Jian et al., 2007; Jian et al., 2007).
In this task, we will use the aforementioned technique involving a spherical deconvolution of analytical Q-ball models to map the local structural anisotropy of the brain tissue and provide at each point of the brain the distribution of fibre orientations as well as all the standard scalar measures like the generalized fractional anisotropy (GFA), the apparent diffusion coefficient (ADC).
We will demonstrate the practical value of this HARDI technique to improve the inference of the cytoarchitectony of the brain stem and the central deep nuclei.
A preliminary work was achieved on this topic and leaded to an accepted abstract at the ISMRM 2009 international conference (see figure 4):
C. Poupon, C. J. Wiggins, M. Descoteaux, T. Feiweier, J-F. Mangin, and D. Le Bihan. "Millimeter analytical Q-ball fiber density function for a better separation of fiber populations at 7T", in proceedings ISMRM 2009