A Novel Approach to Estimate the Intrinsic Dimension of fMRI Data Using Independent Component Analysis

Rajesh Ranjan Nandy ()
Additional contact information
Rajesh Ranjan Nandy: University of North Texas Health Science Center

A chapter in Strategic Management, Decision Theory, and Decision Science, 2021, pp 257-269 from Springer

Abstract: Abstract Estimating the intrinsic dimensionality of data in the presence of noise is an important and well-known problem in signal processing, and there has been extensive work in the literature addressing the problem. However, these existing methods either assume the Gaussian noise in the model to be white or if correlated, require the noise covariance structure to be specified. For fMRI data as well, the estimation of the dimension is an equally important problem and the signal-theoretic methods primarily using information-theoretic approaches are implemented. Unfortunately, for fMRI data, the noise has strong temporal autocorrelation with a complex structure and the covariance structure is unknown. Hence, the traditional methods often provide a biased estimate of true dimension in fMRI data. In this article, independent component analysis (ICA) is used to obtain a crude estimate of the covariance structure of the noise. The estimated structure is then used to whiten the noise, which facilitates a more accurate estimation of the dimension using established information-theoretic methods such as minimum description length (MDL), Bayesian information criteria (BIC) and its Laplace approximation. Results using both real and simulated data are presented, which clearly demonstrate improved performance with the proposed method.

Keywords: fMRI; Temporal autocorrelation; Dimensionality estimation; Independent component analysis (ICA); Minimum description length (MDL); Bayesian information criteria (BIC) (search for similar items in EconPapers)
Date: 2021
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:sprchp:978-981-16-1368-5_16

Ordering information: This item can be ordered from
http://www.springer.com/9789811613685

DOI: 10.1007/978-981-16-1368-5_16

Access Statistics for this chapter

More chapters in Springer Books from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().