Research On Estimation Of Directed Connectivity Among Multichannel EEG Signals

In the fields of neuroscience and biomedical engineering, a great progress of the research on the brain functional connectivity has been made with the rapidly development of the cerebral functional imaging and causality techniques. It provides a groundbreaking way to understand the inner relationship between the cortical functions and its neural activities. The entire cerebral cortex consisting of mass of neural populations can be viewed as a complicated dynamic system. In order to assess the network structure of the mutural effects, various measurements of directional connectivity between multichannel EEG signals have been developed and valicated in recent years.In this paper, several useful estimators and methods are proposed to investiage the dynamic spectral connecitivity between multivariate neural systems and to overcome some drawbacks of the existing methods.First of all, in order to solve the limitation that the conventional Adaptive Directed Function (ADTF) method cannot distinguish instantaneous effect and time-lagged effect between signals, a new extended time-varying multivariate autoregressive model is presented based on a novel Dynamic Causal Ordering (DCO) algorithm. Subsquently, the Instantaneous Effect Factor (IEF) method is defined to track the dynamic instantaneous connectivity while the time-lagged ADTF method is described to measure the time-lagged connectivity between signals. Statistical analysis results suggest that the new DCO algorithm could accurately estimate the causal ordering and give a good estimation of the IEF values in the Gaussian residual conditions. Meanwhile, the time-lagged ADTF approach is also more accurate in estimating the time-lagged dynamic interactions in a complex nervous system after extracting the instantaneous effect.Secondly, the proposed IEF and time-lagged ADTF methods are applied to estimate the dynamic spectral causality on real visual evoked potential (VEP) data in a human subject. Its usefulness in timevariant spectral causality assessment is demonstrated through the contralateral connectivity investigation of cerebral activity during the VEP experiments.Thirdly, to make a multivariate evaluation on the level of weak coupling between signals, the Directed Index (DI) method and Source Index (SI) method are introduced based on the framework of Kuramoto model with noise. The results of the simulation study also indicate their applicability and effectiveness in the identification of source signals. The DI method shows a better performance of source identification than the conventional ADTF method in the weak coupling condition while it also has a pertinent ability of discrimination between the direct and indirect connectivities.Finally, a research on interictal spike data obtained from an epileptic human patient is performed by means of DI and SI methods. Compared to the conclusion obtained by the conventional ADTF method, a different result revealed by DI and ADTF methods suggests that part of the right frontal lobe may have a relatively weak effect to the areas nearby, apart from the distinct and validated influence from the seizure-onset zone of the right temporal lobe for this patient. It is suggested to be an efficient tool to localize the sources of interictal spikes in the patient with epilepsy.