MomayyezSiahkal Parya, Siddiqi Kaleem
Centre for Intelligent Machines, School of Computer Science, McGill University.
Med Image Comput Comput Assist Interv. 2010;13(Pt 1):566-73. doi: 10.1007/978-3-642-15705-9_69.
Diffusion magnetic resonance imaging has led to active research in the analysis of anatomical connectivity in the brain. Many approaches have been proposed to model the diffusion signal and to obtain estimates of fibre tracts. Despite these advances, the question of defining probabilistic connectivity indices which utilize the relevant information in the diffusion MRI signal to indicate connectivity strength, remains largely open. To address this problem we introduce a novel numerical implementation of a stochastic completion field algorithm, which models the diffusion of water molecules in a medium while incorporating the local diffusion MRI data. We show that the approach yields a valid probabilistic estimate of connectivity strength between two seed regions, with experimental results on the MICCAI 2009 Fibre Cup phantom.
扩散磁共振成像已引发了对大脑解剖连接性分析的积极研究。人们提出了许多方法来对扩散信号进行建模并获得纤维束的估计值。尽管取得了这些进展，但定义利用扩散磁共振成像信号中的相关信息来指示连接强度的概率性连接指数这一问题，在很大程度上仍未解决。为解决这一问题，我们引入了一种随机完备场算法的新颖数值实现方法，该方法在纳入局部扩散磁共振成像数据的同时，对水分子在介质中的扩散进行建模。我们表明，该方法能得出两个种子区域之间连接强度的有效概率估计值，并在MICCAI 2009纤维杯模型上给出了实验结果。
