Multitensor approach for analysis and tracking of complex fiber configurations

Abstract

A multidiffusion-tensor model (MDT) is presented containing two anisotropic and one isotropic diffusion tensors. This approach has the ability to detect areas of fiber crossings and resolve the direction of crossing fibers. The mean diffusivity and the ratio of the tensor compartments were merged to one independent parameter by fitting MDT to the diffusion-weighted intensities of a two-point data acquisition scheme. By an F-test between the errors of the standard single diffusion tensor and the more complex MDT, fiber crossings were detected and the more accurate model was chosen voxel by voxel. The performance of crossing detection was compared with the spherical harmonics approach in simulations as well as in vivo. Similar results were found in both methods. The MDT model, however, did not only detect crossings but also yielded the single fiber directions. The FACT algorithm and a probabilistic connectivity algorithm were extended to support the MDT model. Key words: diffusion; DTI; fiber crossings; fiber tracking; multitensorAnisotropy of diffusion in the brain (1) is commonly analyzed by a diffusion tensor (DT) model (2). It is commonly assumed that the principal eigenvector, corresponding to the highest eigenvalue, is aligned along the fiber direction. The direction of this eigenvector has been successfully used for fiber tracking (3-6), which provides unique information about brain structure in vivo (7).Fiber tracking, however, is difficult in brain regions where several fiber bundles are crossing. In these cases, the directions of the principal eigenvectors do not correspond to the fiber directions anymore. The DT in these areas may even have an oblate disk-shape form (8) with an undefined direction of the principle vector. In such cases the common second-order diffusion tensor model is not valid anymore (8 -12). In a voxel with a high amount of isotropic tissue, such as CSF or gray matter, the direction of fibers can also be masked. This could cause errors in the reconstructed fiber trajectories and in connectivity estimation.With high angular resolution DWI measurements (HARDI), i.e., by diffusion encoding along 42-128 optimized directions instead of the six minimum diffusionencoding directions, more accurate and detailed information about diffusion in the brain can be obtained (13-15). Frank applied a spherical harmonics analysis (SH) (9) on HARDI data and was able to detect regions, where common DT models fail and several fiber directions in the same voxel exist. These voxels can be analyzed with a multitensor model (10,16,17) or with a generalized DT (18). Alternatively, Jansons and Alexander (19) introduced a persistent angular structure method (PAS MRI), which employs a maximum entropy method (20). A complex model is required to find multiple fiber directions from HARDI data. Alternatively, more detailed diffusion data measurements, e.g., by Q-space (21,22) or Q-ball (23) experiments, can provide information about fiber directions in crossing regions.The focus of this work was to investigate the p...

Beyond fractional anisotropy: Extraction of bundle-specific structural metrics from crossing fiber models

“…Attempts to solve this problem include restricting the number of tensor components (eg. to two components, Parker et al, 2003, Caan et al, 2010, incorporating physiological constraints (Tuch et al, 2002), reducing the complexity of the model by only allowing identical prolate tensors (Tabelow et al, 2012), stabilizing the problem by using Monte-Carlo algorithms (Kreher et al, 2005), regularizing over a spatial neighborhood (Pasternak et al, 2008, Malcolm et al, 2010, and incorporating other local models to estimate the initial nonlinear optimization of the parameters of the multi-tensor model (Schultz and Kindlmann, 2010).…”

Diffusion magnetic resonance imaging for Brainnetome: A critical review

“…Therefore, it has been generalized into a multi-tensor/ODF streamline by considering the local orientation as the principal, or through the use of multi-tensors [55] , or by considering the maxima of ODFs [20,56,57] . Globally optimized tractography [58][59][60][61][62][63][64][65] was also proposed and has performed better than local tractographic methods.…”

Two-Tensor Fiber Tractography

“…Ref. [15] also uses a multi-tensor approach for analysis and tracking of complex fiber configurations using one isotropic and two anisotropic tensors to model the data.…”