Spatial time series modeling offMRI and EEG data

1)fMRI数据的动态建模与连通性分析本文介绍了一种用于fMRI数据的空间时间序列模型NN-ARX（Nearest Neighbor AutoRegressive Model with eXogenous Variables），并指出了该模型相对于SPM的一些优点。并以视觉刺激的fMRI实验数据为例，说明了该方法在脑电连通性研究中的应用。2）动态脑电逆求解将脑电逆问题转化为多通道脑电时间序列数据的状态空间建模。这里的状态空间是一个动态的主要电流（巨大的维度）体素。提出了一种时空域卡尔曼滤波算法。提出了一种空间卡尔曼滤波算法，并引入了一个逆解作为状态的滤波估计。

1) Dynamic modeling of fMRI data and the connectivity analysisWe introduced a spatial time series model called NN-ARX model (Nearest Neighbor AutoRegressive model with eXogenous variables), for fMRI data.Several advantage of the present model over SPM has been presented. Application of the method to the connectivity study was also shown using the fMRI experiment data of visual stimulus.2) Dynamic EEG inverse solutionEEG inverse problem was reformulated as a state space modeling of multi-channel EEG time series data. Here the state space is a dynamic primary current of (huge dimensional) voxels. A computationally efficient spatial temporal Kalman filtering algorithm was presented. An inverse solution was introduced as a filtered estimate of the state obtained by the proposed spatial Kalman filtering algorithm.