Regresar

P300 Characterization Through Granger Causal Connectivity in the Context of Brain-Computer Interface Technologies

Abstract:

The analysis of connectivity in brain networks has been widely researched and it has been shown that certain cognitive processes require the integration of distributed brain areas. Functional connectivity attempts to statistically quantify the interdependence between these brain areas. In this paper, we propose an analysis of functional connectivity in the Event-Related Potential (ERP) context, more specifically on the P300 component using the Granger Causality measure. To this end, we propose a methodology that consists in quantifying the causality in the P300 and non-P300 signals in the context of Brain-Computer Interfaces (BCIs). Causality is calculated using two approaches: i) using standard electrodes and, ii) using electrodes selected using Bayesian Linear Discriminant Analysis and sequential forward electrode selection (BLDA-FS). Based on this analysis, it is shown that the Granger Causality metric is valid to show a significant connectivity difference between P300 and non-P300 signals. The electrodes selected using BLDA-FS were found to be more discriminative in this regard. Studying functional connectivity using Granger Causality allowed us to identify the changes in connectivity detected during the presence of a target stimulus compared to a non-target stimulus. This additional information about the connectivity differences found can be incorporated as a new feature in further studies, allowing for better detection of the P300 signal and consequently improving the performance of P300-based BCIs.