TY - CHAP
T1 - Weighted complex network based framework for epilepsy detection from EEG signals
AU - Supriya, Supriya
AU - Siuly, Siuly
AU - Wang, Hua
AU - Zhang, Yanchun
N1 - Funding Information:
This work was supported by the National Natural Science Foundation of China (51202186, 51323011) and the Fundamental Research Funds for the Central University (xjj2016039).
Publisher Copyright:
© IOP Publishing Ltd 2020.
PY - 2020/1/1
Y1 - 2020/1/1
N2 - This chapter presents a weighted complex network based framework to identify one of the most challenging brain disorders, namely epilepsy. The pattern of action potentials fluctuates in epileptic disorders, and this can be best intelligible with the assistance of an electroencephalogram (EEG). The EEG is the core authorized biomarker that aids in enhancing the understanding of mental conditions and behaviors, to identify or diagnose any abnormal condition that occurs. The automated diagnosis of epileptic seizure activity using EEG signals is an area of profound ongoing attention in medical science as well as in research disciplines, as the traditional methods of diagnosis of epileptic disorders rely on monotonous visual inspection by highly expert clinicians of long-lasting EEG recordings. The study of complex networks has proved that the underlying dynamics of EEG signals can be best analysed by measuring the strength amongst the nodes of the network, as the topological invariants of the network are closely associated with the underlying dynamics of the EEG. Hence, this chapter introduces an innovative edge-weighted algorithm in the visibility graph for distinguishing epileptic EEG signals from healthy EEG recordings. This research aims to explore the efficacy of the proposed edge weights idea as well as the average weighted degree as efficient network features for identifying epileptic seizure activity using five prevalent machine learning classifiers.
AB - This chapter presents a weighted complex network based framework to identify one of the most challenging brain disorders, namely epilepsy. The pattern of action potentials fluctuates in epileptic disorders, and this can be best intelligible with the assistance of an electroencephalogram (EEG). The EEG is the core authorized biomarker that aids in enhancing the understanding of mental conditions and behaviors, to identify or diagnose any abnormal condition that occurs. The automated diagnosis of epileptic seizure activity using EEG signals is an area of profound ongoing attention in medical science as well as in research disciplines, as the traditional methods of diagnosis of epileptic disorders rely on monotonous visual inspection by highly expert clinicians of long-lasting EEG recordings. The study of complex networks has proved that the underlying dynamics of EEG signals can be best analysed by measuring the strength amongst the nodes of the network, as the topological invariants of the network are closely associated with the underlying dynamics of the EEG. Hence, this chapter introduces an innovative edge-weighted algorithm in the visibility graph for distinguishing epileptic EEG signals from healthy EEG recordings. This research aims to explore the efficacy of the proposed edge weights idea as well as the average weighted degree as efficient network features for identifying epileptic seizure activity using five prevalent machine learning classifiers.
UR - http://www.scopus.com/inward/record.url?scp=85096253940&partnerID=8YFLogxK
M3 - Chapter
AN - SCOPUS:85096253940
SN - 9780750332774
SP - 3-1-3-20
BT - Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
PB - Institute of Physics Publishing
ER -