|
Activity Number:
|
101
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #301800 |
|
Title:
|
Polynomial Spline Independent Component Analysis with Application to fMRI Data
|
|
Author(s):
|
Atsushi Kawaguchi*+ and Young K. Truong and Xuemei Huang
|
|
Companies:
|
The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill
|
|
Address:
|
Department of Biostatistic, Chapel Hill, NC, 27599 ,
|
|
Keywords:
|
Independent Component Analysis ; Spline ; Spatial-Temporal Data ; Functional Magnetic Resonance Imaging
|
|
Abstract:
|
We continue our development of independent component analysis (ICA) using a maximum likelihood approach based on polynomial splines with adaptive knot locations. In a previous study, such approach has produced favorable results based on simulation under the framework of the blind source separation (or temporal ICA), and we also observed that it may not be trivial to apply this method to detect spatial-temporal features from functional magnetic resonance imaging (fMRI) data. In this paper, we incorporate an unsupervised learning method into our algorithm, which yields a significant improvement over existing adaptive methods based on kernel density estimates. The proposed procedure is applied to an fMRI study involving Parkinson's disease patients for identifying regions that are activated by specific experimental tasks.
|
