Wavelet-based Statistical Analysis of fMRI Data

• Activity Number: 47
• Type: Invited
• Date/Time: Sunday, August 7, 2005 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Ivo D. Dinov*+ and John Boscardin and Michael S. Mega and Arthur W. Toga
• Companies: University of California, Los Angeles and University of California, Los Angeles and Neural-Net Research and University of California, Los Angeles
• Keywords: wavelets ; fractals ; fMRI ; neuroimaging ; brain atlas

Abstract

We propose a new method for statistical analysis of functional magnetic resonance imaging (fMRI) data. The discrete wavelet transformation is employed as a tool for efficient and robust signal representation. We use structural MRI and functional fMRI to empirically estimate the distribution of the wavelet coefficients of the data across both individuals and spatial locations. Heavy-tail distributions are then employed to model these data because these signals exhibit slower tail decay than the Gaussian distribution. There are two basic directions we investigate in the first part of this study: Bayesian wavelet-based thresholding scheme, which allows better signal representation, and a family of heavy-tail distributions, which are used as models for the real MRI and fMRI time series data. We discovered Cauchy, Bessel K-Forms, and Pareto distributions provide the most accurate asymptotic models for the distribution of the wavelet coefficients of the data. In the second part of our investigation, we apply this technique to analyze a large fMRI dataset involving repeated presentation of sensory-motor response stimuli in young, elderly, and demented subjects.