Fletcher P Thomas, Tao Ran, Jeong Won-Ki, Whitaker Ross T
School of Computing, University of Utah, Salt Lake City, UT, USA.
Inf Process Med Imaging. 2007;20:346-58. doi: 10.1007/978-3-540-73273-0_29.
In this paper we present a volumetric approach for quantitatively studying white matter connectivity from diffusion tensor magnetic resonance imaging (DT-MRI). The proposed method is based on a minimization of path cost between two regions, defined as the integral of local costs that are derived from the full tensor data along the path. We solve the minimal path problem using a Hamilton-Jacobi formulation of the problem and a new, fast iterative method that computes updates on the propagating front of the cost function at every point. The solutions for the fronts emanating from the two initial regions are combined, giving a voxel-wise connectivity measurement of the optimal paths between the regions that pass through those voxels. The resulting high-connectivity voxels provide a volumetric representation of the white matter pathway between the terminal regions. We quantify the tensor data along these pathways using nonparametric regression of the tensors and of derived measures as a function of path length. In this way we can obtain volumetric measures on white-matter tracts between regions without any explicit integration of tracts. We demonstrate the proposed method on several fiber tracts from DT-MRI data of the normal human brain.
在本文中，我们提出了一种体积法，用于从扩散张量磁共振成像（DT-MRI）定量研究白质连通性。所提出的方法基于两个区域之间路径成本的最小化，路径成本定义为沿路径从完整张量数据导出的局部成本的积分。我们使用该问题的哈密顿-雅可比公式和一种新的快速迭代方法来解决最小路径问题，该方法在成本函数的传播前沿的每个点上计算更新。将从两个初始区域发出的前沿的解相结合，给出了穿过这些体素的区域之间最优路径的逐体素连通性测量。由此产生的高连通性体素提供了终端区域之间白质通路的体积表示。我们使用张量和派生测量值作为路径长度的函数的非参数回归来量化沿这些通路的张量数据。通过这种方式，我们可以在无需对白质束进行任何显式积分的情况下，获得区域之间白质束的体积测量值。我们在来自正常人类大脑DT-MRI数据的几条纤维束上展示了所提出的方法。
