Objective: In studies of brain network, there are different connections. This study aims to explore the differences between the different connection methods. Methods: Using 3T MRI scanner, we collected 22 healthy resting state fMRI data in our experiment. Based on brain activity of motion control, the time courses of prefrontal cortex, motor association cortices, basal ganglia, primary motor cortex, primary sensory cortex, central cerebellum and lateral cerebellum were extracted. Then, pearson correlation, partial correlation, partial least squares, granger causality equation modeling,and structural equation modeling were used to build the connection between the seven brain areas. In order to evaluate the pros and cons of various connection methods, we compared the five models and process signal transduction map of motion control. Results: The results showed that during undirected connected graph, consequence of partial correlation was better, while during directed connected graph, granger causality equation modeling was better matched with the template. Conclusions: In the research of brain network, different connection methods can affect the results. Among the actual research, connection methods should be chosen with the actual experimental conditions and purpose.%目的:在研究脑网络连接过程中,存在不同的连接方式.本文的目的在于探索不同连接方式之间的区别和特点.方法:利用3T磁共振设备,实验当中采集22个健康人静息态功能磁共振数据,依据运动控制过程当中的活动脑区,提取出前额叶皮层、运动联合皮层、基底节、初级运动皮层、初级感觉皮层、小脑中部及小脑侧面区域的时间序列.然后,分别利用Pearson相关、偏相关、偏最小二乘算法、格兰杰因果方程建模、结构方程建模方法来构建上述七个脑区之间的连接.最后,把由五种连接方法建立的结构图与运动控制过程当中的信号传递图做比较,以比较五种不同的连接方法.结果:实验结果表明在无向连接图里面,偏相关显示了较好的结果.在有向连接图里面,格兰杰因果方程建模与模板匹配更好.结论:在脑网络研究当中,不同的连接方法会对实验结果造成不同的影响.实际研究当中,应该结合实际的实验条件和目的,选择合理的连接方法.
展开▼
