Ανάπτυξη μοντέλων μηχανικής μάθησης για την ανίχνευση νευροεκφυλιστικών παθήσεων σε νευροαπεικονιστικά δεδομένα

Παπαζαφειρόπουλος, Αναστάσιος; Papazafeiropoulos, Anastasios

dc.contributor.author	Παπαζαφειρόπουλος, Αναστάσιος	el
dc.contributor.author	Papazafeiropoulos, Anastasios	en
dc.date.accessioned	2024-12-23T07:59:49Z
dc.date.available	2024-12-23T07:59:49Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/60594
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.28290
dc.rights	Αναφορά Δημιουργού - Μη Εμπορική Χρήση - Παρόμοια Διανομή 3.0 Ελλάδα	*
dc.rights.uri	http://creativecommons.org/licenses/by-nc-sa/3.0/gr/	*
dc.subject	Κανονιστική Μοντελοποίηση	el
dc.subject	Νευροεκφυλιστικές Παθήσεις	el
dc.subject	Υπολογιστική Νευροεπιστήμη	el
dc.subject	Αυτοκωδικοποιητές	el
dc.subject	Μηχανική Μάθηση	el
dc.subject	Normative Modeling	en
dc.subject	Neurodegenerative Diseases	en
dc.subject	Computational Neuroscience	en
dc.subject	Autoencoders	en
dc.subject	Machine Learning	en
dc.title	Ανάπτυξη μοντέλων μηχανικής μάθησης για την ανίχνευση νευροεκφυλιστικών παθήσεων σε νευροαπεικονιστικά δεδομένα	el
heal.type	bachelorThesis
heal.secondaryTitle	Κανονιστική μοντελοποίηση με χρήση αυτοκωδικοποιητών	el
heal.classification	Biomedical Engineering	en
heal.language	el
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2024-07
heal.abstract	Detecting brain disorders is critical to understanding and managing changes that affect human consciousness and behavior. Normative modeling offers a breakthrough approach by enabling the detection of pathologies through comparison with normal brain functional patterns. The integration of artificial intelligence and machine learning, in particular autoencoders, greatly enhances the accuracy and efficiency of this technique. This thesis presents the development of two normative modeling frameworks: one using an autoencoder architecture and the other using a variational autoencoder. These models were trained on data from healthy individuals in the UK Biobank. After adjustment for confounders using linear regression and standard scaling, a normative pattern was established. The normative pattern was based on healthy subjects. The models were then tested on semi-synthetic data with simulated atrophy to assess deviations using reconstruction error. Furthermore, the models were evaluated on the ADNI dataset, which includes individuals with normal cognitive function, mild cognitive impairment, and Alzheimer’s disease. The results showed that these models can effectively discriminate between healthy and pathological brain states, paving the way for early diagnosis and treatment of brain disorders. In addition, the models can classify diseases with minimal labeled data for training, providing significant advantages over traditional methods that require extensive labeled datasets.	en
heal.advisorName	Νικήτα, Κωνσταντίνα	el
heal.committeeMemberName	Στάμου, Γεώργιος	el
heal.committeeMemberName	Βουλόδημος, Αθανάσιος	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών. Τομέας Συστημάτων Μετάδοσης Πληροφορίας και Τεχνολογίας Υλικών	el
heal.academicPublisherID	ntua
heal.numberOfPages	81 σ.	el
heal.fullTextAvailability	false