Investigating phase synchronisation in EEG signals for brain
connectivity analysis
Investigating phase synchronisation in EEG signals for brain
connectivity analysis
The brain holds key information regarding the information processing capability of individuals and recent advances in sensor devices and technology have attracted researchers to question the working of this complex organ. It is not only the elusiveness of the brain that has drawn recent research attention but also the claim of doctors that brain function is key in neurological disorders. Disorders like Autism and Attention Deficit Hyperactivity Disorder (ADHD) not to mention other forms of neurobiological diseases have been attributed to disproportionate and disrupted connectivity in the brain. It is envisaged that more accurate and thorough understanding such connectivity can pave the way for medical research of diseases such as these which are deeply rooted to neural level information exchange deficits.
The main objective of this work is to develop an effective means to quantitatively characterise functional connectivity in the brain. Phase synchronisation is reported as the key manifestation of the underlying mechanism of information coupling between different brain regions. This work, therefore first the phase relationships between Electroencephalogram (EEG) signals have been investigated to understand the synchronisation pattern underlying them during the execution of a task. The pursuit to characterise time evolving phase synchrony leads to the identification of the existence of discrete states with quasi-stable phase topography call synchrostates in EEG datasets from range of subjects. These states exhibited switching patterns which were characteristic to the stimuli provided during a cognitive task, specifically in this case face perception tasks. The switching of these states were modelled in a probabilistic framework using a finite Markov model and the stability of the states are represented by the self-transition probabilities.
The degree of phase synchronisation during the existence of each state is then translated into functional connectivity maps and complex network graph measures were applied on it to obtain a set of metrics that quantify the characteristics of such connections formed within the brain. These quantitative brain connectivity measures were used as features to solve a classification problem between autistic and typical children which resulted in an accuracy of 94.7%. The connectivity parameters were then used to characterise behavioural trait scores of anxious children by developing a regression model correlating these to the standardised behavioural scores calculated from questionnaires. Traits like sadness, state anxiety and anger could be modelled effectively using the metrics reported in this study.
This work lays the foundation for further exploration of these quantitative measures for characterising a variety of neurodegenerative diseases and hence may result in a new type of diagnostic process to aid the existing tools available to the clinicians.
Jamal, Wasifa
3f70176e-843e-46b7-8447-4eefaef104f1
February 2015
Jamal, Wasifa
3f70176e-843e-46b7-8447-4eefaef104f1
Maharatna, Koushik
93bef0a2-e011-4622-8c56-5447da4cd5dd
Jamal, Wasifa
(2015)
Investigating phase synchronisation in EEG signals for brain
connectivity analysis.
University of Southampton, Faculty of Physical Sciences and Engineering, Doctoral Thesis, 192pp.
Record type:
Thesis
(Doctoral)
Abstract
The brain holds key information regarding the information processing capability of individuals and recent advances in sensor devices and technology have attracted researchers to question the working of this complex organ. It is not only the elusiveness of the brain that has drawn recent research attention but also the claim of doctors that brain function is key in neurological disorders. Disorders like Autism and Attention Deficit Hyperactivity Disorder (ADHD) not to mention other forms of neurobiological diseases have been attributed to disproportionate and disrupted connectivity in the brain. It is envisaged that more accurate and thorough understanding such connectivity can pave the way for medical research of diseases such as these which are deeply rooted to neural level information exchange deficits.
The main objective of this work is to develop an effective means to quantitatively characterise functional connectivity in the brain. Phase synchronisation is reported as the key manifestation of the underlying mechanism of information coupling between different brain regions. This work, therefore first the phase relationships between Electroencephalogram (EEG) signals have been investigated to understand the synchronisation pattern underlying them during the execution of a task. The pursuit to characterise time evolving phase synchrony leads to the identification of the existence of discrete states with quasi-stable phase topography call synchrostates in EEG datasets from range of subjects. These states exhibited switching patterns which were characteristic to the stimuli provided during a cognitive task, specifically in this case face perception tasks. The switching of these states were modelled in a probabilistic framework using a finite Markov model and the stability of the states are represented by the self-transition probabilities.
The degree of phase synchronisation during the existence of each state is then translated into functional connectivity maps and complex network graph measures were applied on it to obtain a set of metrics that quantify the characteristics of such connections formed within the brain. These quantitative brain connectivity measures were used as features to solve a classification problem between autistic and typical children which resulted in an accuracy of 94.7%. The connectivity parameters were then used to characterise behavioural trait scores of anxious children by developing a regression model correlating these to the standardised behavioural scores calculated from questionnaires. Traits like sadness, state anxiety and anger could be modelled effectively using the metrics reported in this study.
This work lays the foundation for further exploration of these quantitative measures for characterising a variety of neurodegenerative diseases and hence may result in a new type of diagnostic process to aid the existing tools available to the clinicians.
Text
__soton.ac.uk_ude_personalfiles_users_jo1d13_mydesktop_WasifaJamal_Thesis.pdf
- Other
More information
Published date: February 2015
Organisations:
University of Southampton, Electronic & Software Systems
Identifiers
Local EPrints ID: 376540
URI: http://eprints.soton.ac.uk/id/eprint/376540
PURE UUID: 980465f3-5b11-4108-9916-69de423574da
Catalogue record
Date deposited: 03 Jul 2015 16:28
Last modified: 14 Mar 2024 19:45
Export record
Contributors
Author:
Wasifa Jamal
Thesis advisor:
Koushik Maharatna
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics