Advanced

Time Frequency Analysis of EEG Measured When Performing the Flanker Task

Brynolfsson, Johan LU (2012) FMS820 20121
Mathematical Statistics
Abstract
This thesis handles time frequency analysis of EEG signals measured on participants performing the so-called flanker task. The analysis is done mainly using multitapering techniques on the quadratic class. Using multiple orthonormal windows when estimating the spectra of a process, one lowers the variance of estimate.

A class of locally stationary processes (LSP) is presented to use as a model of EEG which can then be used to evaluate the different time-frequency methods that are presented. This LSP contains only one component is used to model only one part of the EEG signal. When analyzing the set of EEG signals of this thesis one is most interested in the so-called N2 event and the model is therefore applied to this event. Having this... (More)
This thesis handles time frequency analysis of EEG signals measured on participants performing the so-called flanker task. The analysis is done mainly using multitapering techniques on the quadratic class. Using multiple orthonormal windows when estimating the spectra of a process, one lowers the variance of estimate.

A class of locally stationary processes (LSP) is presented to use as a model of EEG which can then be used to evaluate the different time-frequency methods that are presented. This LSP contains only one component is used to model only one part of the EEG signal. When analyzing the set of EEG signals of this thesis one is most interested in the so-called N2 event and the model is therefore applied to this event. Having this model one can then find the optimal multitapers in the mean square error sense.

Abstract Different sets of multitapers are used to analyze the time-frequency representation of the EEG-signals. These are evaluated on LSPs where the true spectra are known.

Abstract Spectra are then estimated for the EEG-signals. As there are multiple channels and different methods are used, only a selected set of these spectra are presented here. (Less)
Please use this url to cite or link to this publication:
author
Brynolfsson, Johan LU
supervisor
organization
course
FMS820 20121
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Time Frequency analysis, Locally Stationary processes, Multitaper, EEG
language
English
id
2798050
date added to LUP
2012-06-21 11:54:52
date last changed
2012-06-21 11:54:52
@misc{2798050,
  abstract     = {This thesis handles time frequency analysis of EEG signals measured on participants performing the so-called flanker task. The analysis is done mainly using multitapering techniques on the quadratic class. Using multiple orthonormal windows when estimating the spectra of a process, one lowers the variance of estimate.

A class of locally stationary processes (LSP) is presented to use as a model of EEG which can then be used to evaluate the different time-frequency methods that are presented. This LSP contains only one component is used to model only one part of the EEG signal. When analyzing the set of EEG signals of this thesis one is most interested in the so-called N2 event and the model is therefore applied to this event. Having this model one can then find the optimal multitapers in the mean square error sense.

Abstract Different sets of multitapers are used to analyze the time-frequency representation of the EEG-signals. These are evaluated on LSPs where the true spectra are known.

Abstract Spectra are then estimated for the EEG-signals. As there are multiple channels and different methods are used, only a selected set of these spectra are presented here.},
  author       = {Brynolfsson, Johan},
  keyword      = {Time Frequency analysis,Locally Stationary processes,Multitaper,EEG},
  language     = {eng},
  note         = {Student Paper},
  title        = {Time Frequency Analysis of EEG Measured When Performing the Flanker Task},
  year         = {2012},
}