Functional magnetic resonance imaging (fMRI) is a non-invasive technology that provides high spatial resolution in determining the human brain's responses and measures regional brain activity through metabolic changes in blood oxygen consumption associated with neural activity. Task fMRI provides an opportunity to analyze the working mechanisms of the human brain during specific task performance. Over the past several years, a variety of computational methods have been proposed to decode task fMRI data that can identify brain regions associated with different task stimulations. Despite the advances made by these methods, several limitations exist due to graph representations and graph embeddings transferred from task fMRI signals. In the present study, we proposed an end-to-end graph convolutional network by combining the convolutional neural network with graph representation, with three convolutional layers to classify task fMRI data from the Human Connectome Project (302 participants, 22–35 years of age). One goal of this dissertation was to improve classification performance. We applied four of the most widely used node embedding algorithms—NetMF, RandNE, Node2Vec, and Walklets—to automatically extract the structural properties of the nodes in the brain functional graph, then evaluated the performance of the classification model. The empirical results indicated that the proposed GCN framework accurately identified the brain's state in task fMRI data and achieved comparable macro F1 scores of 0.978 and 0.976 with the NetMF and RandNE embedding methods, respectively. Another goal of the dissertation was to assess the effects of individual differences (i.e., gender and fluid intelligence) on classification performance. We tested the proposed GCN framework on sub-datasets divided according to gender and fluid intelligence. Experimental results indicated significant differences in the classification predictions of gender, but not high/low fluid intelligence fMRI data. Our experiments yielded promising results and demonstrated the superior ability of our GCN in modeling task fMRI data.
| Identifer | oai:union.ndltd.org:ucf.edu/oai:stars.library.ucf.edu:etd2020-2432 |
| Date | 01 January 2022 |
| Creators | Saeidi, Maham |
| Publisher | STARS |
| Source Sets | University of Central Florida |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Electronic Theses and Dissertations, 2020- |
Page generated in 0.0017 seconds