School of Biomedical Engineering, Faculty of Electronic and Electrical Engineering, Dalian University of Technology, 116024, Dalian, China; Faculty of Information Technology, University of Jyväskylä, 40014, Jyväskylä, Finland; Department of Psychiatry, Psychotherapy and Psychosomatics, Medical Faculty, RWTH Aachen University, Pauwelsstraße 30, D-52074, Aachen, Germany.
School of Biomedical Engineering, Faculty of Electronic and Electrical Engineering, Dalian University of Technology, 116024, Dalian, China; Faculty of Information Technology, University of Jyväskylä, 40014, Jyväskylä, Finland.
Neuroimage. 2020 Sep;218:116924. doi: 10.1016/j.neuroimage.2020.116924. Epub 2020 May 20.
Efficient neuronal communication between brain regions through oscillatory synchronization at certain frequencies is necessary for cognition. Such synchronized networks are transient and dynamic, established on the timescale of milliseconds in order to support ongoing cognitive operations. However, few studies characterizing dynamic electrophysiological brain networks have simultaneously accounted for temporal non-stationarity, spectral structure, and spatial properties. Here, we propose an analysis framework for characterizing the large-scale phase-coupling network dynamics during task performance using magnetoencephalography (MEG). We exploit the high spatiotemporal resolution of MEG to measure time-frequency dynamics of connectivity between parcellated brain regions, yielding data in tensor format. We then use a tensor component analysis (TCA)-based procedure to identify the spatio-temporal-spectral modes of covariation among separate regions in the human brain. We validate our pipeline using MEG data recorded during a hand movement task, extracting a transient motor network with beta-dominant spectral mode, which is significantly modulated by the movement task. Next, we apply the proposed pipeline to explore brain networks that support cognitive operations during a working memory task. The derived results demonstrate the temporal formation and dissolution of multiple phase-coupled networks with specific spectral modes, which are associated with face recognition, vision, and movement. The proposed pipeline can characterize the spectro-temporal dynamics of functional connectivity in the brain on the subsecond timescale, commensurate with that of cognitive performance.
大脑区域间通过特定频率的振荡同步进行高效的神经通讯对于认知是必要的。这种同步网络是短暂而动态的，在毫秒级的时间尺度上建立，以支持正在进行的认知操作。然而，很少有研究同时描述动态电生理大脑网络的时间非平稳性、频谱结构和空间特性。在这里，我们提出了一种使用脑磁图（MEG）来描述任务表现期间大规模相位耦合网络动力学的分析框架。我们利用 MEG 的高时空分辨率来测量分区脑区之间连接的时频动力学，产生张量格式的数据。然后，我们使用基于张量成分分析（TCA）的过程来识别人类大脑中不同区域之间的时空频谱变化模式。我们使用在手运动任务期间记录的 MEG 数据验证了我们的管道，提取出具有主导β频带的瞬态运动网络，该网络受到运动任务的显著调制。接下来，我们应用所提出的管道来探索支持工作记忆任务中认知操作的大脑网络。得出的结果表明，具有特定频谱模式的多个相位耦合网络的时间形成和溶解，与面部识别、视觉和运动有关。所提出的管道可以在亚秒级的时间尺度上描述大脑功能连接的谱时动态，与认知表现相当。
