Multi-scale analysis of the dynamics of brain functional connectivity using EEG
Multi-scale analysis of the dynamics of brain functional connectivity using EEG
复制标题
使用脑电图对大脑功能连接的动态进行多尺度分析
DOI：
10.1109/biocas.2016.7833776
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
L. Najafizadeh
中科院分区：
文献类型：
--
作者：
A. Haddad;L. Najafizadeh
本文提出了一种新的方法来研究功能连接的动态在多个时间尺度上从记录通过脑电图（EEG）。利用离散小波变换（DWT）将记录的信号分解为若干个频带，并重构与每个频带相对应的信号（称为子带分量）。使用“源通知分割”技术，然后将子带分量分割成间隔，在该间隔期间，预期活动和功能连接的神经元的潜在集合的空间分布保持准静止。这是通过监视每个子带分量矩阵的运行主导左奇异子空间的跨度随时间的统计显著移位来实现的。最后，对于每个识别的片段，分析源矩阵的功能连接的皮质点的集合。这些合奏本地化，通过相关的时间活动，相应的源矩阵的主导权奇异向量，在段。根据提取的显性右奇异向量的数量，功能上连接的皮层点的多个集合可以被定位为每个段和子带。所提出的方法，然后用于探索视觉古怪的任务，在尺度跨越S-，Q-和A-波段的功能连接的时间演变。结果进行了介绍和讨论。
This paper presents a new approach to investigate the dynamics of functional connectivity at multiple temporal scales from the recordings obtained through electroencephalography (EEG). Discrete wavelet transform (DWT) is utilized to decompose the recorded signals into several frequency bands, and signals corresponding to each frequency band (referred to as subband components) are reconstructed. Using "source-informed segmentation" technique, the subband components are then segmented into intervals during which the spatial distribution of the underlying ensembles of active and functionally connected neurons is expected to stay quasi-stationary. This is achieved through monitoring the running dominant left singular subspace of each subband component matrix for statistically significant shifts in its span over time. Finally, for each identified segment, the source matrix is analyzed for ensembles of functionally connected cortical points. These ensembles are localized through correlating their temporal activity to the dominant right singular vectors of the corresponding source matrix, during the segment. Depending on the number of extracted dominant right singular vectors, multiple ensembles of functionally connected cortical points can be localized per segment and subband. The proposed method is then used to explore the temporal evolution of functional connectivity during a visual oddball task, at scales spanning S-, Q-, and a-bands. Results are presented and discussed.