Novel non-linear approaches to understanding the dynamic brain

: knowledge from rsfMRI and EEG studies

Abstract

Advances in neuroimaging techniques have been critical to identifying new biomarkers for brain diseases. Resting State Functional Magnetic Resonance Imaging (rsfMRI) non-invasively quantifies the Blood Oxygen Level Dependent (BOLD) signal across brain regions with high spatial resolution; whilst temporal resolution of Electroencephalography (EEG) in measuring the brain’s electrical response is unsurpassed. Most of the statistical and machine learning methods used to analyze rsfMRI and EEG data, are static and linear, fail to capture the dynamics and complexity of the brain, and are prone to residual noise. The general goals of this thesis dissertation are i) to provide methodological insight by proposing a statistical method namely point process analysis (PPA) and a machine learning (ML) multiband non-linear EEG method. These methods are especially useful to investigate the brain configuration of older participants and individuals with neurodegenerative diseases, and to predict age and sleep quality; and ii) to share biological insights about synchronization between brain regions (i.e., functional connectivity and dynamic functional connectivity) in different stages of mild cognitive impairment and in Alzheimer’s disease. The findings, reported and discussed in this thesis, open a path for new research ideas such as applying PPA to EEG data, adjusting the non-linear ML algorithm to apply it to rsfMRI and use these methods to better understand other neurological diseases.

Date of Award	13 May 2023
Original language	English
Awarding Institution	Ramon Llull University Universidade Católica Portuguesa Nottingham Trent University
Supervisor	Ignacio Cifre (Supervisor), Patrícia Oliveira-Silva (Co-Supervisor) & Alexander Sumich (Co-Supervisor)