基于格兰杰因果算法的癫痫颅内脑电信号分析

The Electrocorticogram Signal of Seizures Analysis based on Granger Causal Algorithm

利用格兰杰因果关系(Granger Causality,GC)探究癫痫大脑64通道之间的因果关系变化规律。癫痫失神发作前各个通道之间的因果强度为0.01左右,且两两通道之间均有因果关系,这与正常人大脑0.01～0.02的因果强度及方向相一致。失神发作时大脑左颞、顶区与枕区交界处因果信息强度增大到0.2以上,且信息流由此流出,其余脑区因果关系强度也增大到0.05以上。实验结果表明癫痫失神小发作前整个大脑系统的信息流向是平衡的,而癫痫发作时GC强度和方向均发生改变。

In this paper, it uses Granger causality to explore causality variation before, during, and after seizure onset between 64 channels in brain. Before seizure onset, the causality intensity between channels is about 0.01 and it has causal relationship in pairwise channels; it is consistent with the brain of able-bodied person about 0.01-0.02 GC values and information direction. During seizure attack, the GC intensity is above 0.2 and the causal flow is from junction of left temporal region, parietal region and occipital region to the others as well as the GC values of others is above 0.05. The experimental result indicates that it is balanceable about GC direction and intensity of flow before absence epilepsy small attack and it is changed during seizures.