Share Email Print
cover

Proceedings Paper

Bayesian approach for network modeling of brain structural features
Author(s): Anand A. Joshi; Shantanu H. Joshi; Richard M. Leahy; David W. Shattuck; Ivo Dinov; Arthur W. Toga
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Brain connectivity patterns are useful in understanding brain function and organization. Anatomical brain connectivity is largely determined using the physical synaptic connections between neurons. In contrast statistical brain connectivity in a given brain population refers to the interaction and interdependencies of statistics of multitudes of brain features including cortical area, volume, thickness etc. Traditionally, this dependence has been studied by statistical correlations of cortical features. In this paper, we propose the use of Bayesian network modeling for inferring statistical brain connectivity patterns that relate to causal (directed) as well as non-causal (undirected) relationships between cortical surface areas. We argue that for multivariate cortical data, the Bayesian model provides for a more accurate representation by removing the effect of confounding correlations that get introduced due to canonical dependence between the data. Results are presented for a population of 466 brains, where a SEM (structural equation modeling) approach is used to generate a Bayesian network model, as well as a dependency graph for the joint distribution of cortical areas.

Paper Details

Date Published: 9 March 2010
PDF: 8 pages
Proc. SPIE 7626, Medical Imaging 2010: Biomedical Applications in Molecular, Structural, and Functional Imaging, 762607 (9 March 2010); doi: 10.1117/12.844548
Show Author Affiliations
Anand A. Joshi, Lab. of Neuro Imaging, The Univ. of Southern California (United States)
Shantanu H. Joshi, Lab. of Neuro Imaging, The Univ. of Southern California (United States)
Richard M. Leahy, Signal and Image Processing Institute, The Univ. of Southern California (United States)
David W. Shattuck, Lab. of Neuro Imaging, The Univ. of Southern California (United States)
Ivo Dinov, Lab. of Neuro Imaging, The Univ. of Southern California (United States)
Arthur W. Toga, Lab. of Neuro Imaging, The Univ. of Southern California (United States)


Published in SPIE Proceedings Vol. 7626:
Medical Imaging 2010: Biomedical Applications in Molecular, Structural, and Functional Imaging
Robert C. Molthen; John B. Weaver, Editor(s)

© SPIE. Terms of Use
Back to Top