Authors:
Ehsan Chah
1
;
Barry R. Greene
2
;
Geraldine B. Boylan
2
and
Richard B. Reilly
3
Affiliations:
1
UCD School of Electrical, Electronic and Mechanical Engineering, University College Dublin, Ireland
;
2
University College Cork, Ireland
;
3
UCD School of Electrical, Electronic and Mechanical Engineering, University College Dublin; The Cognitive Neurophysiology Laboratory, St Vincent’s Hospital, Ireland
Keyword(s):
Entropy, complexity, neonatal, seizures, EEG.
Related
Ontology
Subjects/Areas/Topics:
Applications and Services
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Computer Vision, Visualization and Computer Graphics
;
Education and Training
;
Medical Image Detection, Acquisition, Analysis and Processing
;
Simulation and Modeling
;
Simulation Tools and Platforms
Abstract:
The performance of three Entropy/complexity measures in detecting EEG seizures in the neonate were investigated in this study. A dataset containing EEG recordings from 11 neonates, with documented electrographic seizures, was employed in this study. Based on patient independent tests Shannon Entropy was found to provide the best in discrimination between seizure and non-seizure EEG in the neonate. Lempel-Ziv complexity and Multi-scale Entropy were second and third respectively, while Sample Entropy did not prove a useful feature for discriminating seizure patterns from non-seizure patterns.