|
Activity Number:
|
52
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #302737 |
|
Title:
|
Multiscale Adaptive Regression Models for Imaging Data
|
|
Author(s):
|
Hongtu Zhu*+ and Yimei Li and Joseph G. Ibrahim and Dinggang Shen
|
|
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 and The University of North Carolina at Chapel Hill
|
|
Address:
|
McGavran Greenberg Hall, CB#7420, Chapel Hill, NC, 27599,
|
|
Keywords:
|
Asymptotic properties ; Multiscale adaptive regression ; Pseudo likelihood ; Wald statistics
|
|
Abstract:
|
We develop a multiscale adaptive regression model (MARM) for spatial and adaptive analysis of imaging data. The MARM fits a regression model based on a pseudo-likelihood function and then constructs a Wald test statistic in each pixel (or voxel) $d$. Then, the MARM successively increases the radius of a spherical neighborhood around $d$ and assign a weight to each voxel $d'$ in the neighborhood of $d$. We construct a weighted pseudo-likelihood function, which combining all data including the weights in a given neighborhood of $d$ with bandwidth $\tilde h$, to obtain adaptive parameter estimators and test statistics. We establish the consistency and asymptotic normality of the adaptive estimators and the asymptotic distribution of the adaptive test statistics for the MARMs. Simulation studies and a real neuroimaging data are used to demonstrate the methodology.
|
