The overall objective of the proposed project is to develop a practical method for mapping white matter pathways over long distances in the brain through regions of complex fiber geometries even in the presence of pathology. Diffusion tensor-based deterministic streamline fiber tracking, using data from Diffusion-Weighted MRI (DW-MRI), can map highly organized white matter pathways, but fails upon encountering fiber crossings and low anisotropy regions typical of neoplasms and white matter disease. A number of functionally important white matter pathways cannot be mapped by conventional methods.
The specific aim of the proposed project is to develop spherical deconvolution with objectively optimized regularization, a method for defining the Fiber Orientation Distribution (FOD), as the basis of probabilistic tracking to define the motor pathway in under one hour with conventional computing resources. The functionally important motor pathway encompasses numerous crossings. Although probabilisitic tracking based on persistent angular structure (PAS) estimation of the FOD has been shown to identify connections to the entire motor pathway, the computational cost of PAS is prohibitive, requiring on the order of 90 cpu-DAYS for an in vivo dataset. Spherical deconvolution with objectively optimized regularization requires two cpu-minutes. However, as PAS-based tracking has identified the entire motor area and has been validated in an animal model, it will serve as the basis for comparison in the absence of a readily accessible gold standard. The proposed project will therefore compare PAS and spherical deconvolution with objectively optimized regularization with regard to their performance in tracking the motor pathway. A 20-processor Linux cluster will enable the PAS calculation and will be used to optimize the spherical deconvolution method. DW-MRI data from 20 healthy subjects will be used to calculate FODs by spherical deconvolution with objectively optimized regularization and PAS as the basis of probabilistic tracking. Primary motor cortex, the seed regions for tracking, will be identified by BOLD-fMRI. Tracks that intersect both bilateral motor cortex regions will be identified as the motor pathway. The goal of this project will have been achieved if a statistically significant correlation between motor pathway identified by each method is found, and the total computation time is under one hour with conventional computing resources. The same analysis will be performed in 10 multiple sclerosis patients as a separate group to evaluate the methodology in the presence of disease. Upon achievement of the overall objective, progress toward a practical method for mapping white matter pathways will have been made. Due to a relative lack of anatomical landmarks on imaging, as compared with gray matter, the function associated with a given region of white matter and damage thereto can be unclear. Development of a more universally applicable method for defining white matter pathways will serve as the basis for improved presurgical planning and better assessment of the importance of injury or repair by therapy to regions of white matter.
White matter in the brain contains functionally important connection pathways. The proposed project aims to develop a practical method for noninvasive mapping of pathways that are difficult or impossible to delineate with current methods, thus extending the utility of such mapping to the entire brain. Improved diagnosis of white matter disease, better assessment of the potential impact of lesions in white matter, and improved presurgical planning may therefore result. ? ? ?
| Taljan, Kyle; McIntyre, Cameron; Sakaie, Ken (2011) Anatomical connectivity between subcortical structures. Brain Connect 1:111-8 |
| Sakaie, Ken E; Lowe, Mark J (2010) Quantitative assessment of motion correction for high angular resolution diffusion imaging. Magn Reson Imaging 28:290-6 |
