National Technical University of Athens (NTUA)
Επεξεργασία και ανάλυση λειτουργικών δικτύων εγκεφάλου επιληπτικών ασθενών σε δεδομένα λειτουργικής απεικόνισης μαγνητικού συντονισμού (fMRI)
Abstract
dc:descriptionOver the past decades, the analysis of brain functional connectivity using resting-state functional Magnetic Resonance Imaging (rs-fMRI) has enriched our knowledge of neurological disorders, including epilepsy. Specifically, the analysis of temporal dependencies between brain regions enables the conceptualization of brain as a complex network, where connections reflect the statistical dependencies of their BOLD signals. Epilepsy has been widely recognized as a brain network disorder, since the pathological activity originating in the epileptogenic zone propagates to anatomically distant regions, yielding to the associated clinical symptomatology. Although EEG-based analysis constitutes the gold standard in clinical practice for the diagnosis of epilepsy, recent studies highlighted the importance of rs-fMRI functional connectivity analysis as a valuable tool for understanding the pathophysiology of the disease. Furthermore, rs-fMRI analysis has significantly improved presurgical evaluation and prediction of postsurgical cognitive outcomes. In the present thesis, we investigate the static and dynamic functional connectivity of brain networks in patients with temporal and extratemporal lobe epilepsy (TLE and ETLE respectively), compared to healthy controls. The purpose of this analysis is to identify novel biomarkers that may deepen our understanding of the underlying mechanisms of epilepsy and support clinical decision-making. Firstly, the analysis is focused on the static directed and undirected functional connectivity utilizing four connectivity methods (Pearson’s correlation, mutual information, Granger causality, transfer entropy). Subsequently, graph-theoretical metrics were applied to the corresponding functional connectivity networks, and the topological features were compared across the two patient groups and healthy controls. Moreover, the detection of the most important regions was conducted using four different centrality measures, to reveal the hubs of each network at a subject level. The hubs for each group were then extracted through an innovative methodological framework based on the hub probability of each region. In the second part of the thesis, we examine the dynamic functional connectivity of the regions included in the Default Mode Network (DMN) utilizing Hidden Markov and Semi-Markov Models (HMM and HSMM respectively). The Markov-based models address the limitations of the traditional sliding-window methods, such as the need to predefine the window length. According to those methods, the timeseries of each region can be modelled as a sequence of latent states, where each state corresponds to a distinct functional connectivity pattern. Following, we compared the features of each state across the three groups. These state features include the total duration spent in each state, the mean dwell time before transitioning to a next state, and the switching rate between states. Furthermore, we evaluate the transition probabilities between the states for each model within each group. The analysis revealed differences between patients with epilepsy and controls, and proved the inability of HMM, compared to HSMM, to capture the dynamics of functional connectivity. The final part of this study focuses on the dynamic analysis of functional connectivity between the DMN and the subcortical regions, utilizing the Hidden Semi-Markov Models. Instead of applying the model to all the concatenated data, the hidden states were separately evaluated for each group. Subsequently, the graph theoretical measures were estimated for each extracted state to assess the difference among the identified motifs. Moreover, the hubs of each state were detected for each group utilizing four centrality measures. Consistent with the static analysis, the dynamic states exhibited a distinct hub distribution between epilepsy patients and healthy controls.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Amoiridou, Dimitra
- Αμοιρίδου, Δήμητρα
Subjects
dc:subject × 17- Λειτουργική απεικόνιση μαγνητικού συντονισμού
- Επιληψία
- Λειτουργικά δίκτυα του εγκεφάλου
- Λειτουργική συνδεσιμότητα
- Θεωρία γράφων
- Κρυφό (Ημι-) Μαρκοβιανό Μοντέλο
- Resting state fMRI
- Epilepsy
- Functional brain networks
- Functional connectivity
- Graph theory
- Hidden (Semi-) Markov models
- Βιοϊατρική Μηχανική
- Επιστήμες Μηχανικού και Τεχνολογία
- Medical Engineering
- Engineering and Technology
- Biomedical Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/60608