| Popis: |
We deal with the characterization of the functional brain connectivity of subjects under different mental states using theoretical graph analysis. This characterization is important since getting a description of the brain functional connectivity can identify brain areas related with specialized functions. This problem it is not trivial because the information of an electroencephalogram (EEG) is highly redundant and requires methods that simplify its analysis. In this work we introduce a method based on a graph, which we call the Common Edge Graph (CEG), constructed from EEG data for functional brain characterization. The CEG results from the intersection of a collection of graphs, one graph for each subject representing the most similar electrode pair signals. In contrast to other methods, our approach gets a connective representation considering the edges which are invariant in all subjects under the same mental state and looks for the same edges in test subjects discarding irrelevant information in order to simplify the resulting graph interpretation. Then, using this new representation of brain connectivity, we show that it is possible to recognize the most prominent connective differences between subjects under distinct mental states. We evaluate the performance of our approach in three experiments: open vs closed eyes under resting state, hand motion vs open eyes under resting state and foot vs hand motion. |
