Murdoch University Research Repository

Welcome to the Murdoch University Research Repository

The Murdoch University Research Repository is an open access digital collection of research
created by Murdoch University staff, researchers and postgraduate students.

Learn more

Identification of EEG signal patterns between adults with dyslexia and normal controls

Perera, Biyagama (2017) Identification of EEG signal patterns between adults with dyslexia and normal controls. PhD thesis, Murdoch University.

PDF - Whole Thesis
Download (7MB) | Preview


Electroencephalography (EEG) is one of the most useful techniques used to represent behaviours of the brain and helps explore valuable insights through the measurement of brain electrical activity. Hence, it plays a vital role in detecting neurological disorders such as epilepsy. Dyslexia is a hidden learning disability with a neurological origin affecting a significant amount of the world population. Studies show unique brain structures and behaviours in individuals with dyslexia and these variations have become more evident with the use of techniques such as EEG, Functional Magnetic Resonance Imaging (fMRI), Magnetoencephalography (MEG) and Positron Emission Tomography (PET).

In this thesis, we are particularly interested in discussing the use of EEG to explore unique brain activities of adults with dyslexia. We attempt to discover unique EEG signal patterns between adults with dyslexia compared to normal controls while performing tasks that are more challenging for individuals with dyslexia. These tasks include real--‐word reading, nonsense--‐ word reading, passage reading, Rapid Automatized Naming (RAN), writing, typing, browsing the web, table interpretation and typing of random numbers. Each participant was instructed to perform these specific tasks while staying seated in front of a computer screen with the EEG headset setup on his or her head. The EEG signals captured during these tasks were examined using a machine learning classification framework, which includes signal preprocessing, frequency sub--‐band decomposition, feature extraction, classification and verification. Cubic Support Vector Machine (CSVM) classifiers were developed for separate brain regions of each specified task in order to determine the optimal brain regions and EEG sensors that produce the most unique EEG signal patterns between the two groups.

The research revealed that adults with dyslexia generated unique EEG signal patterns compared to normal controls while performing the specific tasks. One of the vital discoveries of this research was that the nonsense--‐words classifiers produced higher Validation Accuracies (VA) compared to real--‐ words classifiers, confirming difficulties in phonological decoding skills seen in individuals with dyslexia are reflected in the EEG signal patterns, which was detected in the left parieto--‐occipital. It was also uncovered that all three reading tasks showed the same optimal brain region, and RAN which is known to have a relationship to reading also showed optimal performance in an overlapping region, demonstrating the likelihood that the association between reading and RAN reflects in the EEG signal patterns. Finally, we were able to discover brain regions that produced exclusive EEG signal patterns between the two groups that have not been reported before for writing, typing, web browsing, table interpretation and typing of random numbers.

Item Type: Thesis (PhD)
Murdoch Affiliation(s): School of Engineering and Information Technology
United Nations SDGs: Goal 3: Good Health and Well-Being
Supervisor(s): Shiratuddin, Fairuz and Wong, Kevin
Item Control Page Item Control Page


Downloads per month over past year