TY - CHAP
T1 - Characterizing EEG Electrodes in Directed Functional Brain Networks Using Normalized Transfer Entropy and PageRank
AU - Suresh, Kaushik
AU - Ramasamy, Vijayalakshmi
AU - Daniel, Ronnie
AU - Chandra, Sushil
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - Over the years, cognitive research has been an active and evolving field, where non-invasive techniques like Electroencephalogram (EEG) play a dominant role in the study of brain functions. The electrical signals recorded from the brain using multi-channel EEG are used in a wide range of applications, making it possible to understand the concept of cognition better. It is crucial to study the non-linear and dynamic electrical signals generated from the brain in order to understand the behavior of each brain region. This study provides a better understanding of cognition by using page ranking (a widely used algorithm to rank a website in a network of sites using the principle of eigenvector centrality) of the EEG electrodes. Based on the performance in the short-term memory task called Corsi Block-tapping task (CBTT), the participants are classified into two groups, viz. good and poor performers. The directed Functional Brain Networks (FBNs) are constructed using Normalized Transfer Entropy (NTE) by considering EEG electrodes as nodes in the network, information flow between pairs of nodes as edges, and the NTE values of connectivity as edge weights. The NTE values computed during the performance of the CBTT task of the participants are compared with the baseline data for both good and poor performers. The weighted page rank algorithm is used to compute the ranks of the electrodes in terms of the cognitive load measured using NTE values of the different brain regions. The status of each electrode at the two groups is identified using the Reduction and Increase in Consistency (RIC) value. Therefore, the RIC value serves as an indicator of the decrease/increase/constant in the rank of an electrode during the performance of the CBTT task compared to that during the baseline. A user-defined Observation Phase Value (OPV) number of the top-ranked electrodes is used to analyze the cognitive processes within the groups (using three different arbitrary OPV values of 10, 20, and 30). Based on the ranks of the electrodes and an OPV value of 20, the common electrodes during the baseline and CBTT task activity are classified using four different categories of occurrences (100, 81–99, 61–80 and 50–60%) for all the good and poor performers of CBTT respectively. The empirical analysis helps characterize the EEG electrodes into different categories based on cognitive activities using efficient computational techniques. The inferences made from such an analysis play a significant role in understanding the cognitive behavior of human brain networks using the directional flow of information during cognitive load-based tasks such as short-term memory CBTT task.
AB - Over the years, cognitive research has been an active and evolving field, where non-invasive techniques like Electroencephalogram (EEG) play a dominant role in the study of brain functions. The electrical signals recorded from the brain using multi-channel EEG are used in a wide range of applications, making it possible to understand the concept of cognition better. It is crucial to study the non-linear and dynamic electrical signals generated from the brain in order to understand the behavior of each brain region. This study provides a better understanding of cognition by using page ranking (a widely used algorithm to rank a website in a network of sites using the principle of eigenvector centrality) of the EEG electrodes. Based on the performance in the short-term memory task called Corsi Block-tapping task (CBTT), the participants are classified into two groups, viz. good and poor performers. The directed Functional Brain Networks (FBNs) are constructed using Normalized Transfer Entropy (NTE) by considering EEG electrodes as nodes in the network, information flow between pairs of nodes as edges, and the NTE values of connectivity as edge weights. The NTE values computed during the performance of the CBTT task of the participants are compared with the baseline data for both good and poor performers. The weighted page rank algorithm is used to compute the ranks of the electrodes in terms of the cognitive load measured using NTE values of the different brain regions. The status of each electrode at the two groups is identified using the Reduction and Increase in Consistency (RIC) value. Therefore, the RIC value serves as an indicator of the decrease/increase/constant in the rank of an electrode during the performance of the CBTT task compared to that during the baseline. A user-defined Observation Phase Value (OPV) number of the top-ranked electrodes is used to analyze the cognitive processes within the groups (using three different arbitrary OPV values of 10, 20, and 30). Based on the ranks of the electrodes and an OPV value of 20, the common electrodes during the baseline and CBTT task activity are classified using four different categories of occurrences (100, 81–99, 61–80 and 50–60%) for all the good and poor performers of CBTT respectively. The empirical analysis helps characterize the EEG electrodes into different categories based on cognitive activities using efficient computational techniques. The inferences made from such an analysis play a significant role in understanding the cognitive behavior of human brain networks using the directional flow of information during cognitive load-based tasks such as short-term memory CBTT task.
KW - Directed information flow
KW - Electroencephalography (EEG)
KW - Functional brain networks (FBN)
KW - PageRank
KW - Transfer entropy
UR - http://www.scopus.com/inward/record.url?scp=85115629869&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-79161-2_2
DO - 10.1007/978-3-030-79161-2_2
M3 - Chapter
AN - SCOPUS:85115629869
T3 - Intelligent Systems Reference Library
SP - 27
EP - 49
BT - Intelligent Systems Reference Library
PB - Springer Science and Business Media Deutschland GmbH
ER -