Variability of Independent Components in functional Magnetic Resonance Imaging

Reference:

Jarkko Ylipaavalniemi. Variability of independent components in functional magnetic resonance imaging. Master's thesis, Helsinki University of Technology, Espoo, Finland, March 2005.

Abstract:

Independent component analysis (ICA) has been widely adopted as a powerful data-driven signal processing technique. But, even though it has been shown to be helpful in many fields, such as, biomedical systems, telecommunication, finance and natural image processing, there remains problems in its wide adoption. One concern is that solutions found with ICA algorithms tend to change slightly each time analysis is done, raising serious questions about the reliability of those solutions. This behavior stems from the stochastic nature of the data and ICA algorithms, and affects many other blind source separation (BSS) algorithms as well. This thesis presents a method to analyze the consistency of the solutions. It is also shown how to exploit the variability to gain additional information on the found solutions. The method is based on clustering solutions from multiple runs of bootstrapped ICA. Its usefulness is tested with a real functional magnetic resonance imaging (fMRI) experiment, involving auditory stimulus, where several independent components are truly consistent. Additionally, the information acquired with the method helps in analyzing the underlying phenomena of the less consistent ones.

Keywords:

independent component analysis, functional magnetic resonance imaging

Suggested BibTeX entry:

@mastersthesis{JYlipaavalniemi-2005-MSCTHESIS,
    address = {Espoo, Finland},
    author = {Jarkko Ylipaavalniemi},
    month = {March},
    school = {Helsinki University of Technology},
    title = {Variability of Independent Components in functional Magnetic Resonance Imaging},
    year = {2005},
}