Multiscale adaptive regression models for neuroimaging data

Yimei Li, Hongtu Zhu, Dinggang Shen, Weili Lin, John H. Gilmore, Joseph G. Ibrahim

Research output: Contribution to journalArticle

43 Citations (Scopus)

Abstract

Neuroimaging studies aim to analyse imaging data with complex spatial patterns in a large number of locations (called voxels) on a two-dimensional surface or in a three-dimensional volume. Conventional analyses of imaging data include two sequential steps: spatially smoothing imaging data and then independently fitting a statistical model at each voxel. However, conventional analyses suffer from the same amount of smoothing throughout the whole image, the arbitrary choice of extent of smoothing and low statistical power in detecting spatial patterns. We propose a multiscale adaptive regression model to integrate the propagation- separation approach with statistical modelling at each voxel for spatial and adaptive analysis of neuroimaging data from multiple subjects. The multiscale adaptive regression model has three features: being spatial, being hierarchical and being adaptive. We use a multiscale adaptive estimation and testing procedure to utilize imaging observations from the neighbouring voxels of the current voxel to calculate parameter estimates and test statistics adaptively. Theoretically, we establish consistency and asymptotic normality of the adaptive parameter estimates and the asymptotic distribution of the adaptive test statistics. Our simulation studies and real data analysis confirm that the multiscale adaptive regression model significantly outperforms conventional analyses of imaging data.

Original languageEnglish
Pages (from-to)559-578
Number of pages20
JournalJournal of the Royal Statistical Society. Series B: Statistical Methodology
Volume73
Issue number4
DOIs
Publication statusPublished - 2011 Sep 1
Externally publishedYes

Fingerprint

Neuroimaging
Voxel
Regression Model
Imaging
Smoothing
Spatial Pattern
Test Statistic
Adaptive Testing
Adaptive Test
Adaptive Estimation
Statistical Power
Statistical Modeling
Asymptotic Normality
Estimate
Statistical Model
Asymptotic distribution
Regression model
Data analysis
Integrate
Simulation Study

Keywords

  • Kernel
  • Multiscale adaptive regression
  • Neuroimaging data
  • Propagation-separation
  • Smoothing
  • Sphere
  • Test statistics

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Multiscale adaptive regression models for neuroimaging data. / Li, Yimei; Zhu, Hongtu; Shen, Dinggang; Lin, Weili; Gilmore, John H.; Ibrahim, Joseph G.

In: Journal of the Royal Statistical Society. Series B: Statistical Methodology, Vol. 73, No. 4, 01.09.2011, p. 559-578.

Research output: Contribution to journalArticle

Li, Yimei ; Zhu, Hongtu ; Shen, Dinggang ; Lin, Weili ; Gilmore, John H. ; Ibrahim, Joseph G. / Multiscale adaptive regression models for neuroimaging data. In: Journal of the Royal Statistical Society. Series B: Statistical Methodology. 2011 ; Vol. 73, No. 4. pp. 559-578.
@article{f537b1783515441d9459161ba78430a9,
title = "Multiscale adaptive regression models for neuroimaging data",
abstract = "Neuroimaging studies aim to analyse imaging data with complex spatial patterns in a large number of locations (called voxels) on a two-dimensional surface or in a three-dimensional volume. Conventional analyses of imaging data include two sequential steps: spatially smoothing imaging data and then independently fitting a statistical model at each voxel. However, conventional analyses suffer from the same amount of smoothing throughout the whole image, the arbitrary choice of extent of smoothing and low statistical power in detecting spatial patterns. We propose a multiscale adaptive regression model to integrate the propagation- separation approach with statistical modelling at each voxel for spatial and adaptive analysis of neuroimaging data from multiple subjects. The multiscale adaptive regression model has three features: being spatial, being hierarchical and being adaptive. We use a multiscale adaptive estimation and testing procedure to utilize imaging observations from the neighbouring voxels of the current voxel to calculate parameter estimates and test statistics adaptively. Theoretically, we establish consistency and asymptotic normality of the adaptive parameter estimates and the asymptotic distribution of the adaptive test statistics. Our simulation studies and real data analysis confirm that the multiscale adaptive regression model significantly outperforms conventional analyses of imaging data.",
keywords = "Kernel, Multiscale adaptive regression, Neuroimaging data, Propagation-separation, Smoothing, Sphere, Test statistics",
author = "Yimei Li and Hongtu Zhu and Dinggang Shen and Weili Lin and Gilmore, {John H.} and Ibrahim, {Joseph G.}",
year = "2011",
month = "9",
day = "1",
doi = "10.1111/j.1467-9868.2010.00767.x",
language = "English",
volume = "73",
pages = "559--578",
journal = "Journal of the Royal Statistical Society. Series B: Statistical Methodology",
issn = "1369-7412",
publisher = "Wiley-Blackwell",
number = "4",

}

TY - JOUR

T1 - Multiscale adaptive regression models for neuroimaging data

AU - Li, Yimei

AU - Zhu, Hongtu

AU - Shen, Dinggang

AU - Lin, Weili

AU - Gilmore, John H.

AU - Ibrahim, Joseph G.

PY - 2011/9/1

Y1 - 2011/9/1

N2 - Neuroimaging studies aim to analyse imaging data with complex spatial patterns in a large number of locations (called voxels) on a two-dimensional surface or in a three-dimensional volume. Conventional analyses of imaging data include two sequential steps: spatially smoothing imaging data and then independently fitting a statistical model at each voxel. However, conventional analyses suffer from the same amount of smoothing throughout the whole image, the arbitrary choice of extent of smoothing and low statistical power in detecting spatial patterns. We propose a multiscale adaptive regression model to integrate the propagation- separation approach with statistical modelling at each voxel for spatial and adaptive analysis of neuroimaging data from multiple subjects. The multiscale adaptive regression model has three features: being spatial, being hierarchical and being adaptive. We use a multiscale adaptive estimation and testing procedure to utilize imaging observations from the neighbouring voxels of the current voxel to calculate parameter estimates and test statistics adaptively. Theoretically, we establish consistency and asymptotic normality of the adaptive parameter estimates and the asymptotic distribution of the adaptive test statistics. Our simulation studies and real data analysis confirm that the multiscale adaptive regression model significantly outperforms conventional analyses of imaging data.

AB - Neuroimaging studies aim to analyse imaging data with complex spatial patterns in a large number of locations (called voxels) on a two-dimensional surface or in a three-dimensional volume. Conventional analyses of imaging data include two sequential steps: spatially smoothing imaging data and then independently fitting a statistical model at each voxel. However, conventional analyses suffer from the same amount of smoothing throughout the whole image, the arbitrary choice of extent of smoothing and low statistical power in detecting spatial patterns. We propose a multiscale adaptive regression model to integrate the propagation- separation approach with statistical modelling at each voxel for spatial and adaptive analysis of neuroimaging data from multiple subjects. The multiscale adaptive regression model has three features: being spatial, being hierarchical and being adaptive. We use a multiscale adaptive estimation and testing procedure to utilize imaging observations from the neighbouring voxels of the current voxel to calculate parameter estimates and test statistics adaptively. Theoretically, we establish consistency and asymptotic normality of the adaptive parameter estimates and the asymptotic distribution of the adaptive test statistics. Our simulation studies and real data analysis confirm that the multiscale adaptive regression model significantly outperforms conventional analyses of imaging data.

KW - Kernel

KW - Multiscale adaptive regression

KW - Neuroimaging data

KW - Propagation-separation

KW - Smoothing

KW - Sphere

KW - Test statistics

UR - http://www.scopus.com/inward/record.url?scp=79961044921&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79961044921&partnerID=8YFLogxK

U2 - 10.1111/j.1467-9868.2010.00767.x

DO - 10.1111/j.1467-9868.2010.00767.x

M3 - Article

AN - SCOPUS:79961044921

VL - 73

SP - 559

EP - 578

JO - Journal of the Royal Statistical Society. Series B: Statistical Methodology

JF - Journal of the Royal Statistical Society. Series B: Statistical Methodology

SN - 1369-7412

IS - 4

ER -