Information dynamics view of brain processing function

We present a methodology for the analysis of electromagnetic (EM) brain signals. In a dynamical systems framework we assume that the measured electroencephalogram (EEG) and the magnetoencephalogram (MEG) are generated by the non-linear interaction of a few degrees of freedom. Within this framework, we then construct an embedding matrix, which consists of a series of consecutive delay vectors. The embedding matrix describes a trajectory on the Euclidean manifold recreating the unobservable system manifold, which is assumed to be generating the measured data. The embedding matrix can be used to quantify system complexity, which changes with the changes in brain-'state'. To this end, we use measures of entropy and Fisher's information measure to track changes in complexity of the system over time. It is also possible to perform Independent Component Analysis on the embedding matrix to decompose the single channel recording into a set of underlying independent components. The independent components are treated as a convenient expansion basis and subjective methods are used to identify components of interest relevant to the application at hand. The method is applied to just single channels of both EEG and MEG recordings and is shown to give intuitive and meaningful results in a neurophysiological setting.