Use this URL to cite or link to this record in EThOS:
Title: Connectivity analysis from EEG phase synchronisation in emotional BCI
Author: Santamaria, Lorena
ISNI:       0000 0004 6423 1676
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
A Brain Computer Interface (BCI) is a device that uses the brain activity of the user as an input to the system to select the desired output on a computer, giving the person a different pathway to establish communications with the surrounding environment. There are many types and uses of BCIs. They can be defined by which technique is used to record the brain activity of the user and which variety of stimuli is used to trigger a consistent response from the user, following the signal processing methodology selected to produce a response on the computer. Each one of the selected choices will determine the reliability and efficiency of the BCI system. However, even with this flexibility, the performance of BCI systems used for assistive technology or rehabilitation processes still remains behind other systems and the percentage of people unable to use one of these systems remains too high. The main objective of this thesis is to improve the classification performance and reliability of the current electroencephalogram (EEG) based BCI systems. Firstly, a novel paradigm based on emotional faces is used with the aim of enhancing a stronger response from the user, therefore a higher amplitude of brain activity. Two types of emotional faces have been used during this work. Initially, emotional schematic faces or emotions were used. Posteriorly, human emotional faces were introduced into the experiments. Additionally, the evolution of the phase synchronisation over time is studied to achieve a deeper understanding of the latent communication mechanisms of the different parts of the human brain. Wavelet families and their ability to retain temporal and frequency information simultaneously have been used to study the phase relationships between the EEG signals when a specific task is being performed. This study has led to the identification of a reduced number of discrete states with a quasi-stable phase synchronisation of the order of milliseconds, named synchrostates. Those synchrostates present switching patterns over time, clearly distinctive for each one of the tasks performed by the user. In order to establish a classification protocol the temporal stability of each task-specific synchrostate was studied by means of the synchronisation index and posteriorly translated into connectivity network maps based on graph theory. From this connectivity network, a series of connectivity metrics was obtained and used to feed a variety of classification algorithms. This process led to accuracies of 83% for a two-tasks classification problem and rose to a 93% averaged accuracy for a four-tasks problem.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: QP Physiology ; TK Electrical engineering. Electronics Nuclear engineering