Surgical Planning Laboratory - Brigham & Women's Hospital - Boston, Massachusetts USA - a teaching affiliate of Harvard Medical School

Surgical Planning Laboratory

The Publication Database hosted by SPL

All Publications | Upload | Advanced Search | Gallery View | Download Statistics | Help | Import | Log in

SGPP: Spatial Gaussian Predictive Process Models for Neuroimaging Data

Institution:
1Department of Biostatistics, St. Jude Children's Research Hospital, Memphis, TN, USA.
2Department of Psychiatry, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA.
3Department of Biostatistics, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA.
Publisher:
Elsevier Science
Publication Date:
Apr-2014
Journal:
Neuroimage
Volume Number:
89
Pages:
70-80
Citation:
Neuroimage. 2014 Apr 1;89:70-80.
PubMed ID:
24269800
PMCID:
PMC4134945
Keywords:
Cokriging, Functional Principal Component Analysis, Missing Data, Prediction, Simultaneous Autoregressive Model, Spatial Gaussian Predictive Process
Appears in Collections:
NA-MIC
Sponsors:
R21 AG033387/AG/NIA NIH HHS/United States
R01 HD053000/HD/NICHD NIH HHS/United States
P50 MH064065/MH/NIMH NIH HHS/United States
R01 MH070890/MH/NIMH NIH HHS/United States
R01 MH086633/MH/NIMH NIH HHS/United States
R01 MH091645/MH/NIMH NIH HHS/United States
P01 CA142538/CA/NCI NIH HHS/United States
P30 HD003110/HD/NICHD NIH HHS/United States
R01 MH091645/MH/NIMH NIH HHS/United States
R01 ES017240/ES/NIEHS NIH HHS/United States
UL1 RR025747/RR/NCRR NIH HHS/United States
U54 EB005149/EB/NIBIB NIH HHS/United States
Generated Citation:
Hyun J.W., Li Y., Gilmore J.H., Lu Z., Styner M., Zhu H. SGPP: Spatial Gaussian Predictive Process Models for Neuroimaging Data. Neuroimage. 2014 Apr 1;89:70-80. PMID: 24269800. PMCID: PMC4134945.
Downloaded: 628 times. [view map]
Paper: Download, View online
Export citation:
Google Scholar: link

The aim of this paper is to develop a spatial Gaussian predictive process (SGPP) framework for accurately predicting neuroimaging data by using a set of covariates of interest, such as age and diagnostic status, and an existing neuroimaging data set. To achieve a better prediction, we not only delineate spatial association between neuroimaging data and covariates, but also explicitly model spatial dependence in neuroimaging data. The SGPP model uses a functional principal component model to capture medium-to-long-range (or global) spatial dependence, while SGPP uses a multivariate simultaneous autoregressive model to capture short-range (or local) spatial dependence as well as cross-correlations of different imaging modalities. We propose a three-stage estimation procedure to simultaneously estimate varying regression coefficients across voxels and the global and local spatial dependence structures. Furthermore, we develop a predictive method to use the spatial correlations as well as the cross-correlations by employing a cokriging technique, which can be useful for the imputation of missing imaging data. Simulation studies and real data analysis are used to evaluate the prediction accuracy of SGPP and show that SGPP significantly outperforms several competing methods, such as voxel-wise linear model, in prediction. Although we focus on the morphometric variation of lateral ventricle surfaces in a clinical study of neurodevelopment, it is expected that SGPP is applicable to other imaging modalities and features.

Additional Material
1 File (109.926kB)
Hyan-Neuroimage2014-fig6.jpg (109.926kB)