Login | Register

Electroencephalogram spindle detection from simultaneous Electroencephalogram (EEG) / functional Magnetic Resonance Imaging (fMRI) data for a sleep deprivation study

Title:

Electroencephalogram spindle detection from simultaneous Electroencephalogram (EEG) / functional Magnetic Resonance Imaging (fMRI) data for a sleep deprivation study

Jegou, Aude (2018) Electroencephalogram spindle detection from simultaneous Electroencephalogram (EEG) / functional Magnetic Resonance Imaging (fMRI) data for a sleep deprivation study. Masters thesis, Concordia University.

[img]
Preview
Text (application/pdf)
Jegou_MSc_S2018.pdf - Accepted Version
Available under License Spectrum Terms of Access.
6MB

Abstract

Sleep deprivation is an important and common problem with many consequences for mental and physical health. A sleep deprivation study was conducted to better understand the effects of insufficient sleep on cognitive functions and recovery. Studying specific spontaneous sleep waves occurring during the recovery sleep after sleep deprivation allows us to better understand the brain mechanism during sleep and its importance. The focus of my Master project is the detection of these sleep waves on scalp Electro-EncephaloGraphy (EEG) data, especially spindles, using different automatic detection algorithms and comparing them to a visual detection. Automated detection of these typical EEG discharges is challenging because many artifacts occur when EEG data are recorded within the Magnetic Resonance Imaging (MRI) scanner, so in presence of a large magnetic field. EEG data should be carefully processed from MRI related artifacts before considering detection of sleep specific discharges. In the second section of the thesis, we investigated within the whole brain the Blood Oxygen Level Dependent (BOLD) responses, measured with functional MRI, to identify brain areas involved in the generation of sleep spindles detected from scalp EEG. Preliminary results showed that automated detection was not accurate enough, because of residual artifact and other consequences of preprocessing. Moreover, visual detection is limited by the complexity of the EEG data. Therefore, there are expected improvements in post-review of the automated detection or investigating the optimal parameters for automated methods. The first BOLD response maps showed similarity between automated and visual detection and also with a study conducted by (Manuel Schabus et al. 2007). To conclude, further investigation on automated methods must be performed to find the best compromise between visual detection, automated detection, and post-review of automated detection.

Divisions:Concordia University > Faculty of Arts and Science > Physics
Item Type:Thesis (Masters)
Authors:Jegou, Aude
Institution:Concordia University
Degree Name:M. Sc.
Program:Physics
Date:April 2018
Thesis Supervisor(s):Grova, Christophe and Dang-Vu, Thanh
ID Code:983635
Deposited By: Aude Jegou
Deposited On:11 Jun 2018 03:26
Last Modified:11 Jun 2018 03:26
All items in Spectrum are protected by copyright, with all rights reserved. The use of items is governed by Spectrum's terms of access.

Repository Staff Only: item control page

Downloads per month over past year

Back to top Back to top