Significance of Event Related Causality (ERC) in Neural Networks
3400 N. Charles Street
Jump to:
Bio
“Significance of Event Related Causality (ERC) in Neural Networks”
Dr. Anna Korzeniewska obtained MS in Physics with concentration in Medical Physics from University of Warsaw, Poland and PhD in Biological Sciences with concentration in Neurophysiology from Nencki Institute of Experimental Biology, Polish Academy of Sciences. Since 2004 she works at Johns Hopkins’ Epilepsy Center. Her research interest is focused on the dynamics of causal interactions among functional and pathological neural networks.
Abstract
“Significance of Event Related Causality (ERC) in Neural Networks”
Neural activity is propagated across large-scale cortical networks on very brief time scales. Studying such transient and complex systems calls for a short time-window on one hand, and a great extent of recording sites in the brain, on the other. These demands are not easily satisfied, as short time intervals do not provide enough data-points to model the dynamics of large-scale brain networks. The limitation can be overcome by using multiple realizations of the same process, but the price to be paid is that traditional statistical methods cannot be used to assess the significance of event-related changes in the estimated dynamics of the system. To obtain statistical confidence of the dynamics of neural interactions among large-scale networks revealed by event-related causality (ERC), we propose using the variance of a two-dimensional moving average. We also propose a criterion for the two-dimensional model selection, which combines the difference between the smooth estimator and the real values with the confidence interval. We show that this estimator is efficient, stable, and ensures precise embedding of statistical significance in two-dimensional (time-frequency) space. Here, we show that the method can be used to investigate information flow among eloquent network, to provide a guidance for epileptic surgery.
