Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The Statistical Analysis of Functional MRI DataMultivariate Approaches

The Statistical Analysis of Functional MRI Data: Multivariate Approaches In this chapter we look at fMRI data from the multivariate perspectives of component and correlation analyses. The former include principal components analysis (PCA) and independent components analysis (ICA); the latter in- clude canonical correlation analysis and maximum correlation analysis. ICA is by far the most popular of these methods. All of the procedures of this chapter share the feature that they are “data driven” rather than “model” or “hypothesis driven.” The implication is that the researcher does not need to specify a priori all the possible effects and behaviors of interest; indeed, the components that are produced as a result of the various decompositions will often lend themselves to unexpected interpretations. For instance, in addition to components that are task-related, with associated time courses that follow the experimental paradigm (and which could be predicted in advance), the methods can discover components that correspond to transient effects, and even some that don’t relate specifically to the task, but are consistently found across subjects or cluster spatially, indicating their “veracity” as elements of interest (see, for instance, Calhoun et al. 2001a). This seems to be especially true of ICA, and is perhaps one explanation for its popularity (Hu et al., http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

The Statistical Analysis of Functional MRI DataMultivariate Approaches

Download PDF
24 pages

Loading next page...
 
/lp/springer-journals/the-statistical-analysis-of-functional-mri-data-multivariate-Bj5VokezUx
Publisher
Springer New York
Copyright
© Springer-Verlag New York 2008
ISBN
978-0-387-78190-7
Pages
1 –24
DOI
10.1007/978-0-387-78191-4_7
Publisher site
See Chapter on Publisher Site

Abstract

In this chapter we look at fMRI data from the multivariate perspectives of component and correlation analyses. The former include principal components analysis (PCA) and independent components analysis (ICA); the latter in- clude canonical correlation analysis and maximum correlation analysis. ICA is by far the most popular of these methods. All of the procedures of this chapter share the feature that they are “data driven” rather than “model” or “hypothesis driven.” The implication is that the researcher does not need to specify a priori all the possible effects and behaviors of interest; indeed, the components that are produced as a result of the various decompositions will often lend themselves to unexpected interpretations. For instance, in addition to components that are task-related, with associated time courses that follow the experimental paradigm (and which could be predicted in advance), the methods can discover components that correspond to transient effects, and even some that don’t relate specifically to the task, but are consistently found across subjects or cluster spatially, indicating their “veracity” as elements of interest (see, for instance, Calhoun et al. 2001a). This seems to be especially true of ICA, and is perhaps one explanation for its popularity (Hu et al.,

Published: Jun 7, 2008

Keywords: Principal Component Analysis; Independent Component Analysis; Canonical Correlation; Canonical Correlation Analysis; fMRI Data

There are no references for this article.