Dynamic analysis of resting-state EEG reveals altered brain states and functional network transition
Poster No:
287
Submission Type:
Abstract Submission
Authors:
Marjorie Metzger1, Stefan Dukic2, Roisin McMackin1, Eileen Giglia1, Matthew Mitchell1, Emmet Costello1, Saroj Bista1, Colm PEELO1, Yasmine Tadjine1, Vladyslav Sirenko3, Mark Heverin1, Peter Bede3, Muthuraman Muthuraman4, Orla Hardiman5, Bahman Nasseroleslami5
Institutions:
1Trinity College Dublin, Dublin, Dublin, 2UMC, Utrecht, Netherlands, 3Trinity College Dublin, Dublin, Dublin , 4Johannes Gutenberg Hospital, Dublin, Dublin , 5Trinity College Dublin, Dublin, Ireland
First Author:
Co-Author(s):
Introduction:
Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder affecting motor neurons. Its multi-faceted nature encompasses a wide spectrum of symptoms, including muscle weakness and eventual paralysis, but also respiratory and cognitive symptoms. A promising approach for studying this complex condition at the level of underlying networks, involved leveraging the spectral power and functional connectivity of resting-state EEG. These measures can discriminate the abnormal network function in individuals with ALS versus healthy controls (HC) [1]–[3]. Based on our recent findings in resting-state EEG microstates [4], we hypothesize that dynamic analysis of recurring patterns in resting state EEG based on source-level measures of spectral power and coherence can further elucidate the altered network function in ALS and in providing reliable domain-specific indicators of impairment in ALS.
Objectives: To identify transient brain states associated with specific functional networks, using high-density resting-state EEG, as well as to characterise the spatio-spectral alterations in these brain states and their dynamics in individuals with ALS.
Objectives: To identify transient brain states associated with specific functional networks, using high-density resting-state EEG, as well as to characterise the spatio-spectral alterations in these brain states and their dynamics in individuals with ALS.
Methods:
High-density resting-state EEG data were recorded from 99 individuals with ALS and 78 HC. To identify transient and recurrent brain states, we applied a time-delay embedded Hidden Markov Model to source-reconstructed resting-state EEG data (1-45Hz) [5]. The model was trained to convert source-reconstructed time courses into a sequence of functional networks characterised by spectral power and coherence. Subsequently, we employed non-negative matrix factorisation to break down the spectral measures for each state into four components, roughly corresponding to the frequency bands commonly used in electrophysiology (delta/theta, alpha, beta, gamma) [5]. Properties of the sequences of brain states were then analysed to determine their recurrence frequency, duration, and transition patterns. For each state, nonparametric statistical analyses, based on Area Under the Curve (AUC) [as test-statistic] and empirical Bayesian Inference (EBI) [6] [for multivariate inference], were conducted to evaluate the significance of differences in spectral measures between the ALS and HC groups. Furthermore, we explored correlations with clinical evaluations of functional, cognitive and behavioural impairments to assess how well these brain states might serve as domain-specific indicators of impairment.
·Figure 1: Brain states analysis pipeline. Description of the method used to compute brain state timecourses from EEG data.
Results:
Twelve brain states were identified with distinct patterns of spectral power and coherence for individuals with ALS and HC. States in HC had longer intervals, indicating a greater number of timepoints between state visits, for state 1 (q=0.004, AUC = 0.63, 1-β_0.05=0.76) and state 10 (q=0.001, AUC = 0.75, 1-β_0.05=0.88). States 1, 3, 7 and 9 showed significant association with behavioural decline (as reported using the Beaumont Behavioural Inventory [7]; ∣r_s∣ > 0.25, q < 0.03, 1-β_0.05 > 0.65), while state 5 showed association with fluency decline (evaluated using the Edinburgh Cognitive and Behavioural ALS scale [8]; r_s = -0.3 , q = 0.004, 1-β_0.05 = 0.83). States 1, 7 and 10 were characterised by frontal lobe activation (spectral power higher than the average within the state), while state 3 exhibited activation in the sensorimotor network. State 5 highest spectral power was in the supplementary motor area, a region which as been linked not only with motor planning but also with speech [9].
·Figure 2. Z-scores of the spectral power in the twelve states resulting from the TDE-HMM model in the overall group (HC and ALS groups combined).
Conclusions:
This study demonstrates altered dynamics of functional networks in ALS. The use of dynamical analysis of spectral brain states provides insights into transitions between functional networks. The findings confirm the potential of spectral resting-state EEG measures as potential multi-domain quantitative marker of abnormal changes in brain networks in ALS. This study also paves the way for investigating the relationship between alterations in EEG signals and specific functional domains in ALS.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1
Higher Cognitive Functions:
Higher Cognitive Functions Other
Modeling and Analysis Methods:
EEG/MEG Modeling and Analysis 2
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping
Keywords:
Cognition
Data analysis
Degenerative Disease
Electroencephaolography (EEG)
Motor
Movement Disorder
Somatosensory
Statistical Methods
1|2Indicates the priority used for review
Provide references using author date format
[2] S. Dukic et al., “Patterned functional network disruption in amyotrophic lateral sclerosis,” Hum. Brain Mapp., vol. 0, no. 0, Jul. 2019, doi: 10.1002/hbm.24740.
[3] S. Dukic et al., “Resting-state EEG reveals four subphenotypes of amyotrophic lateral sclerosis,” Brain, vol. 145, no. 2, pp. 621–631, Apr. 2022, doi: 10.1093/brain/awab322.
[4] M. Metzger et al., “Functional network dynamics revealed by EEG microstates reflect cognitive decline in amyotrophic lateral sclerosis,” Oct. 2023, doi: 10.1002/hbm.26536.
[5] D. Vidaurre et al., “Spontaneous cortical activity transiently organises into frequency specific phase-coupling networks,” Nat. Commun., vol. 9, no. 1, Art. no. 1, Jul. 2018, doi: 10.1038/s41467-018-05316-z.
[6] B. Nasseroleslami, “An Implementation of Empirical Bayesian Inference and Non-Null Bootstrapping for Threshold Selection and Power Estimation in Multiple and Single Statistical Testing,” bioRxiv, p. 342964, Jun. 2018, doi: 10.1101/342964.
[7] M. Elamin et al., “Identifying behavioural changes in ALS: Validation of the Beaumont Behavioural Inventory (BBI),” Amyotroph. Lateral Scler. Front. Degener., vol. 18, no. 1–2, pp. 68–73, Jan. 2017, doi: 10.1080/21678421.2016.1248976.
[8] M. Pinto-Grau et al., “Screening for cognitive dysfunction in ALS: validation of the Edinburgh Cognitive and Behavioural ALS Screen (ECAS) using age and education adjusted normative data,” Amyotroph. Lateral Scler. Front. Degener., vol. 18, no. 1–2, pp. 99–106, Jan. 2017, doi: 10.1080/21678421.2016.1249887.
[9] F.-X. Alario, H. Chainay, S. Lehericy, and L. Cohen, “The role of the supplementary motor area (SMA) in word production,” Brain Res., vol. 1076, no. 1, pp. 129–143, Mar. 2006, doi: 10.1016/j.brainres.2005.11.104.