The Application Of Causality In Detecting Brain Effective Connectivity

The brain is a very complicated system, in which neurons, neuronal populations orbrain regions link each other forming into complex structural networks, and completebrain various functions through interactions between distinct units. The brain is an organthat serves as the center of the nervous system. A major aspect of the complexity ofnervous system relates to their intricate morphology, especially the interconnectivity oftheir neuronal processing elements. Neural connectivity patterns have long attracted theattention of neuroanatomists and play crucial roles in determining the functionalproperties of neurons and neuronal systems. Brain connectivity pattern refers tostructural connectivity (anatomical connectivity), functional connectivity or effectiveconnectivity. Anatomical connectivity refers to a network of physical or structuralconnections linking sets of neurons or neuronal elements. Functional connectivitycaptures interrelation in dynamic physiology activity from statistical independencebetween distributed and often spatially remote neuronal units and describes whetherbrain regions are connected, the strength of the connections. Effective connectivitydescribes networks of directional effects of one neural element over another. In order toexplore effective brain connectivity, we study the motor imagery tasks and applycausality methods to mainly analyze the interactions of three brain regions: C3, Cz andC4.We deeply study bivariate Granger causality and New causality respectively fromtime domain and frequency domain in this paper, meanwhile compare the two causalitymethod and detect the shortcomings of Granger Causality, the advantages of Newcausality. Also the research visualizes the application of causality method and thedifferences of the two methods through concrete example models.Because motor imagery task is the research object, we need to deeply understandthe relevant knowledge. Electroencephalography (EEG), Brain-computer interface(BCI), and the mechanism and characteristics of motor imagery are learned in this paper.The research pays attention to Mu rhythm in motor imagery as the causality shows inMu rhythm. The data sets which are studied are provided by international competitionso the conclusion of the research owns authenticity.The important content in the paper is to apply causality methods to analyze thedata sets recorded when does motor imagery tasks to detect causal influences between each two channels. The channels are C3, Cz and C4. At first, we reveal the features ofMu rhythm by power spectrum analysis. Then for the causality results, we focus on theperformance in Mu rhythm, such as the comparison between two causality values orwhether peak value appears in Mu rhythm. We draw the following conclusions from ouranalysis on27data sets recorded from9volunteers:1) During left and right hand motor imagery, the causality of Cz on C3/C4islarger than that of C3/C4on Cz, that is, there is strong directional connectivityfrom Cz to C3/C4.2) During left hand motor imagery, the causality of C4on C3in Mu rhythm islarger than that of C3on C4, in other words there is strong causal influencefrom C4to C3.3) During right hand motor imagery, the causality of C3on C4in Mu rhythm islarger than that of C4on C3, in other words there is strong causal influencefrom C3to C4.Comparing the results by GC method with the results by NC method, we get NCmethod can better reveal real causal influence among Cz, C3and C4three regions inMu rhythm than GC method during MI.In this paper, we provide new evidence to support that new causality method isbetter than Granger causality to reveal true causality. We deeply believe new causalitymethod will replace Granger causality to reflect causality and will be widely used inmany areas such as economy, engineering and neurosciences.