Document Type

Dissertation

Degree

Doctor of Philosophy (PhD)

Major/Program

Electrical Engineering

First Advisor's Name

Dr. Mercedes Cabrerizo

First Advisor's Committee Title

Committee Chair

Second Advisor's Name

Dr. Malek Adjouadi

Second Advisor's Committee Title

Committee member

Third Advisor's Name

Dr. Armando Barreto

Third Advisor's Committee Title

Committee member

Fourth Advisor's Name

Dr. Jean Andrian

Fourth Advisor's Committee Title

Committee member

Fifth Advisor's Name

Dr. Naphtali David Rishe

Fifth Advisor's Committee Title

Committee member

Keywords

Epilepsy, Brain Connectivity Analysis, Effective Connectivity, Machine Learning

Date of Defense

11-9-2018

Abstract

This dissertation introduces a novel approach at gauging patterns of informa- tion flow using brain connectivity analysis and partial directed coherence (PDC) in epilepsy. The main objective of this dissertation is to assess the key characteristics that delineate neural activities obtained from patients with epilepsy, considering both focal and generalized seizures. The use of PDC analysis is noteworthy as it es- timates the intensity and direction of propagation from neural activities generated in the cerebral cortex, and it ascertains the coefficients as weighted measures in formulating the multivariate autoregressive model (MVAR). The PDC is used here as a feature extraction method for recorded scalp electroencephalograms (EEG) as means to examine the interictal epileptiform discharges (IEDs) and reflect the phys- iological changes of brain activity during interictal periods. Two experiments were set up to investigate the epileptic data by using the PDC concept.

For the investigation of IEDs data (interictal spike (IS), spike and slow wave com- plex (SSC), and repetitive spikes and slow wave complex (RSS)), the PDC analysis estimates the intensity and direction of propagation from neural activities gener- ated in the cerebral cortex, and analyzes the coefficients obtained from employing MVAR. Features extracted by using PDC were transformed into adjacency matrices

using surrogate data analysis and were classified by using the multilayer Perceptron (MLP) neural network. The classification results yielded a high accuracy and pre- cision number.

The second experiment introduces the investigation of intensity (or strength) of information flow. The inflow activity deemed significant and flowing from other regions into a specific region together with the outflow activity emanating from one region and spreading into other regions were calculated based on the PDC results and were quantified by the defined regions of interest. Three groups were considered for this study, the control population, patients with focal epilepsy, and patients with generalized epilepsy. A significant difference in inflow and outflow validated by the nonparametric Kruskal-Wallis test was observed for these groups.

By taking advantage of directionality of brain connectivity and by extracting the intensity of information flow, specific patterns in different brain regions of interest between each data group can be revealed. This is rather important as researchers could then associate such patterns in context to the 3D source localization where seizures are thought to emanate in focal epilepsy. This research endeavor, given its generalized construct, can extend for the study of other neurological and neurode- generative disorders such as Parkinson, depression, Alzheimers disease, and mental illness.

Identifier

FIDC007001

Available for download on Wednesday, October 16, 2019

Share

COinS
 

Rights Statement

Rights Statement

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).