Extracting Robust Biomarkers From Multichannel EEG Time Series Using Nonlinear Dimensionality Reduction Applied to Ordinal Pattern Statistics and Spectral Quantities

Inga Kottlarz; Sebastian Berg; Diana Toscano-Tejeida; Iris Steinmann; Mathias Bähr; Stefan Luther; Melanie Wilke; Ulrich Parlitz; Alexander Schlemmer

doi:10.3389/fphys.2020.614565

Extracting Robust Biomarkers From Multichannel EEG Time Series Using Nonlinear Dimensionality Reduction Applied to Ordinal Pattern Statistics and Spectral Quantities

Front Physiol. 2021 Feb 1:11:614565. doi: 10.3389/fphys.2020.614565. eCollection 2020.

Authors

Inga Kottlarz^{1

2}, Sebastian Berg¹, Diana Toscano-Tejeida³, Iris Steinmann³, Mathias Bähr⁴, Stefan Luther^{1

5

6}, Melanie Wilke^{3

7}, Ulrich Parlitz^{1

2

6}, Alexander Schlemmer^{1

6}

Affiliations

¹ Max Planck Institute for Dynamics and Self-Organization, Göttingen, Germany.
² Institute for the Dynamics of Complex Systems, Georg-August-Universität Göttingen, Göttingen, Germany.
³ Department of Cognitive Neurology, University Medical Center Göttingen, Göttingen, Germany.
⁴ Department of Neurology, University Medical Center Göttingen, Göttingen, Germany.
⁵ Institute of Pharmacology and Toxicology, University Medical Center Göttingen, Göttingen, Germany.
⁶ German Center for Cardiovascular Research (DZHK), Partner Site Göttingen, Göttingen, Germany.
⁷ German Primate Center, Leibniz Institute for Primate Research, Göttingen, Germany.

Abstract

In this study, ordinal pattern analysis and classical frequency-based EEG analysis methods are used to differentiate between EEGs of different age groups as well as individuals. As characteristic features, functional connectivity as well as single-channel measures in both the time and frequency domain are considered. We compare the separation power of each feature set after nonlinear dimensionality reduction using t-distributed stochastic neighbor embedding and demonstrate that ordinal pattern-based measures yield results comparable to frequency-based measures applied to preprocessed data, and outperform them if applied to raw data. Our analysis yields no significant differences in performance between single-channel features and functional connectivity features regarding the question of age group separation.

Keywords: EEG - Electroencephalogram; biomarkers; coherence; functional connectivity; mutual information; nonlinear dimensionality reduction; ordinal pattern statistics; t-SNE (t-distributed stochastic neighbor embedding).