National Technical University of Athens (NTUA)
Ανάλυση εγκεφαλικής δραστηριότητας και ταξινόμηση νευροφυσιολογικών καταστάσεων με χρήση δεδομένων ηλεκτροεγκεφαλογραφήματος και μαγνητικής τομογραφίας
Abstract
dc:descriptionThis PhD thesis addresses the analysis of brain activity and the classification of neurophysiological states in healthy individuals and patients, exploiting electroencephalography (EEG) and diffusion tensor imaging (DTI) data and employing methods from statistics, machine learning, and deep learning. Based on these techniques, the work aims to discriminate between different states of brain function and to identify significant features, focusing on cognitive processes related to memory, mental fatigue, real and imagined movement, as well as on overall brain connectivity. The objectives of the thesis include the investigation of the applicability of user-friendly cognitive training protocols, the utilisation of brain–computer interfaces for the study and support of motor functions, and the analysis of brain reorganisation at the structural level due to disease and therapeutic interventions. Within this framework, the contribution of portable and low-complexity EEG systems to the assessment of cognitive functions and memory processes is investigated. Through an experimental cognitive training protocol based on the distinction between automatic and controlled memory processes, both behavioural indices and electrophysiological features of induced activity (ERS/ERD) are analysed. The results demonstrate that even single-channel portable EEG devices can provide meaningful information for the interpretation of cognitive changes, enhancing behavioural assessment. Subsequently, the thesis focuses on the study of mental fatigue due to sleep deprivation, exploiting multichannel EEG data and functional connectivity indices combined with the selection of interpretable features based on machine learning methods. The analysis revealed critical differentiations in interactions among brain regions, corresponding to disruptions in executive control and working memory networks. In parallel, a multifaceted analysis of real and imagined lower-limb movement data is conducted, acquired through a specialised experimental protocol aimed at investigating the feasibility of designing and training a brain-computer interface (BCI) system. In this context, correlation-based techniques are applied to compare real and imagined movement, while the role of guidance stimuli (visual or auditory according to the acquisition protocol) is also examined with respect to brain activity differentiation and classification performance based on machine learning. Furthermore, deep learning techniques (with the application of a convolutional neural network) are employed on these data for the classification of real movements, achieving high performance and highlighting their potential use in rehabilitation applications. Finally, the thesis extends to a clinical context by examining structural alterations in brain connectivity in patients with small-cell lung cancer (SCLC) after chemotherapy. Through the analysis of DTI data and the application of machine learning techniques combined with a framework for selecting robust and interpretable features, a small subset of anatomical connections was identified that distinguished patients from healthy controls with high accuracy, highlighting effects on brain functions closely related to cognitive performance. Overall, the thesis demonstrates the potential for interpretable classification and analysis of brain states based on electrophysiological and neuroimaging data, in combination with behavioural cognitive metrics and the application of diverse protocols covering cognitive training, mental fatigue analysis, motor function analysis, and the study of the effects of disease and therapeutic approaches.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Miloulis, Stavros-Theofanis
- Μηλούλης, Σταύρος-Θεοφάνης
Subjects
dc:subject × 33- Εγκέφαλος
- Γνωσιακές διεργασίες
- Εγκεφαλική δραστηριότητα
- Νευροπλαστικότητα
- Εγκεφαλικά δίκτυα
- Διεπαφή εγκεφάλου-υπολογιστή
- Ηλεκτροεγκεφαλογραφία
- Μαγνητική τομογραφία
- Απεικόνιση τανυστή διάχυσης
- Τεχνητή νοημοσύνη
- Μηχανική μάθηση
- Βαθιά μάθηση
- Brain
- Cognitive processes
- Brain activity
- Neuroplasticity
- Brain networks
- Brain-Computer Interface
- Electroencephalography
- Magnetic resonance imaging
- Diffusion tensor imaging
- Artificial intelligence
- Machine learning
- Deep learning
- Βιοϊατρική Μηχανική
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Ηλεκτρική και Ηλεκτρονική μηχανική
- Medical Engineering
- Engineering and Technology
- Biomedical Engineering
- Electrical Engineering, Electronic Engineering, Information Engineering
- Electrical and Electronic Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/60885