University of Houston

The study of the brain and its complex small and large scale networks has led us to realize the importance of neural integration in information processing. Our ability to examine the functional dynamics of brain activity has transformed with new methods for measuring the magnetic flux associated with changing cortical activity. Neuroconnectivity measurement techniques have evolved to measure the direction and strength of connections within neural networks.

In this dissertation, we introduce an implementation of Granger causality that allows the calculation of causal interconnectivity between many sources simultaneously. This algorithm, Dynamic Autoregressive Neuromagnetic Causal Imaging (DANCI), considers the various interrelationships between multiple sources dynamically, thereby providing a true measure of causal connectivity while allowing the indirect influence of other sources to be taken into account. We compared DANCI to the traditional Granger causality method and evaluated six multivariate causality connectivity measures using simulation and clinical data. Performance was measured by the accuracy of predicted connections in the simulated network. We concluded that DANCI allows the accurate, fast calculation of causal interaction in complex networks. In addition, the DANCI algorithm was applied to clinical magnetoencephalographic data to successfully determine the differences in brain connectivity between normal and dyslexic subjects.