Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/D3EF2B48-D234-4DDF-9177-30FE9333A998 | ||
| Year | 2019 | ||
| Type of Item | Master Thesis | ||
| License |
|
||
| Bibliographic Citation | Paris Karakasis, "Machine Learning Techniques with Applications in Biomedical Signal Processing", Master Thesis, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2019 https://doi.org/10.26233/heallink.tuc.84115 | ||
| Appears in Collections |
Functional magnetic resonance imaging (fMRI) is one of the most popular methods for studying the human brain. The purpose of task-related fMRI data analysis is to determine which brain areas are activated when a specific task is performed, based on the (Blood Oxygen Level Dependent) signal analysis. Background BOLD signal reflects systematic fluctuations in regional brain activity that are attributed to the existence of resting-state brain networks. A wide range of unsupervised multivariate statistical methods is being increasingly employed in fMRI data analysis. The main goal of these methods is to extract information from a dataset, often with no prior knowledge of the experimental conditions. Generalized canonical correlation analysis (gCCA) is a well known statistical method, that can be considered as the method for the estimation of a linear subspace, which is ”common” to multiple sets of random vectors. We propose a new fMRI data generating model, which takes into consideration the existence of common task-related and rest-related components. Moreover, we estimate the task-related components via gCCA. We extensively test our theoretical results using both artificial and real-world fMRI data. We observe that our experimental findings corroborate our theoretical results, rendering our approach a very good candidate for multi-subject task-related fMRI processing.
Copyright © DIAS 2013
