A0657
Title: Measuring cross-channel information transfer in the frequency domain through spectral transfer entropy
Authors: Raphael Huser - King Abdullah University of Science and Technology (Saudi Arabia)
Paolo Victor Redondo - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Hernando Ombao - KAUST (Saudi Arabia)
Abstract: Brain connectivity reflects how different regions of the brain interact during the performance of a cognitive task. Studying brain signals, such as electroencephalograms (EEG), may be explored via transfer entropy (TE). This information-theoretic causal measure covers any form of relationship (beyond linear) between variables. To improve the utility of TE, a novel methodology is proposed to capture cross-channel information transfer in the frequency domain. A new causal measure, spectral transfer entropy (STE), is introduced to quantify the magnitude and direction of information flow from a frequency-band oscillation of a channel to an oscillation of another channel. In contrast with previous works on TE in the frequency domain, this method is differentiated by considering an extreme value perspective that employs the maximum magnitude of filtered series within time blocks. The main advantages of this approach are that it is robust to the issues of linear filtering and allows adjustments for multiple comparisons to control family-wise error rates. Another novel contribution is a simple yet efficient estimation method based on the combination of vine copulas and extreme value theory that enables estimates to capture zero (boundary point) is illustrated without the need for bias adjustments. Lastly, the advantage of our measure through numerical experiments and interesting and novel findings are provided in the analysis of EEG recordings linked to a visual task.
