Eigenconnectivities of dynamic functional networks: Consistency across subjects

• Published in: Conference record - Asilomar Conference on Signals, Systems, & Computers pp. 620 - 623
• Main Authors: Leonardi, Nora, Van De Ville, Dimitri
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2014
• Subjects: canonical correlation analysis; Correlation; dynamic functional connectivity; Eigenvalues and eigenfunctions; functional MRI; Head; Magnetic resonance imaging; Matrix decomposition; Principal component analysis; Time series analysis
• ISSN: 1058-6393
• DOI: 10.1109/ACSSC.2014.7094520

Functional connectivity (FC) measured using fMRI has provided significant insights into brain function. However, increasing evidence points towards continuously fluctuating FC across the duration of a scan. Using unsupervised learning techniques, reproducible patterns of dynamic FC (dFC) have been revealed. In particular, based on principal component analysis, it has recently been proposed to represent dFC as a linear combination of multiple "eigenconnectivities". These group-level results were obtained by concatenating all subjects' timecourses of dFC. Here we investigate the consistency of these results by introducing a subject-level and group-level PCA and comparing the results with those obtained by concatenation.