Non-parametric Bayesian graph models reveal community structure in resting state fMRI

Kasper Winther Andersen, Kristoffer H. Madsen, Hartwig Roman Siebner, Mikkel Nørgaard Schmidt, Morten Mørup, Lars Kai Hansen

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

Modeling of resting state functional magnetic resonance imaging (rs-fMRI) data using network models is of increasing interest. It is often desirable to group nodes into clusters to interpret the communication patterns between nodes. In this study we consider three different nonparametric Bayesian models for node clustering in complex networks. In particular, we test their ability to predict unseen data and their ability to reproduce clustering across datasets. The three generative models considered are the Infinite Relational Model (IRM), Bayesian Community Detection (BCD), and the Infinite Diagonal Model (IDM). The models define probabilities of generating links within and between clusters and the difference between the models lies in the restrictions they impose upon the between-cluster link probabilities. IRM is the most flexible model with no restrictions on the probabilities of links between clusters. BCD restricts the between-cluster link probabilities to be strictly lower than within-cluster link probabilities to conform to the community structure typically seen in social networks. IDM only models a single between-cluster link probability, which can be interpreted as a background noise probability. These probabilistic models are compared against three other approaches for node clustering, namely Infomap, Louvain modularity, and hierarchical clustering. Using 3 different datasets comprising healthy volunteers' rs-fMRI we found that the BCD model was in general the most predictive and reproducible model. This suggests that rs-fMRI data exhibits community structure and furthermore points to the significance of modeling heterogeneous between-cluster link probabilities.
Original languageEnglish
JournalNeuroImage
Volume100
Pages (from-to)301-315
ISSN1053-8119
DOIs
Publication statusPublished - 2014

Keywords

  • Complex network
  • Graph theory
  • Infinite Relational Model
  • Bayesian Community Detection
  • Resting state fMRI

Cite this

@article{42db11aacbdc4d2aa525ba61b7328c33,
title = "Non-parametric Bayesian graph models reveal community structure in resting state fMRI",
abstract = "Modeling of resting state functional magnetic resonance imaging (rs-fMRI) data using network models is of increasing interest. It is often desirable to group nodes into clusters to interpret the communication patterns between nodes. In this study we consider three different nonparametric Bayesian models for node clustering in complex networks. In particular, we test their ability to predict unseen data and their ability to reproduce clustering across datasets. The three generative models considered are the Infinite Relational Model (IRM), Bayesian Community Detection (BCD), and the Infinite Diagonal Model (IDM). The models define probabilities of generating links within and between clusters and the difference between the models lies in the restrictions they impose upon the between-cluster link probabilities. IRM is the most flexible model with no restrictions on the probabilities of links between clusters. BCD restricts the between-cluster link probabilities to be strictly lower than within-cluster link probabilities to conform to the community structure typically seen in social networks. IDM only models a single between-cluster link probability, which can be interpreted as a background noise probability. These probabilistic models are compared against three other approaches for node clustering, namely Infomap, Louvain modularity, and hierarchical clustering. Using 3 different datasets comprising healthy volunteers' rs-fMRI we found that the BCD model was in general the most predictive and reproducible model. This suggests that rs-fMRI data exhibits community structure and furthermore points to the significance of modeling heterogeneous between-cluster link probabilities.",
keywords = "Complex network, Graph theory, Infinite Relational Model, Bayesian Community Detection, Resting state fMRI",
author = "Andersen, {Kasper Winther} and Madsen, {Kristoffer H.} and Siebner, {Hartwig Roman} and Schmidt, {Mikkel N{\o}rgaard} and Morten M{\o}rup and Hansen, {Lars Kai}",
year = "2014",
doi = "10.1016/j.neuroimage.2014.05.083",
language = "English",
volume = "100",
pages = "301--315",
journal = "NeuroImage",
issn = "1053-8119",
publisher = "Elsevier",

}

Non-parametric Bayesian graph models reveal community structure in resting state fMRI. / Andersen, Kasper Winther; Madsen, Kristoffer H.; Siebner, Hartwig Roman; Schmidt, Mikkel Nørgaard; Mørup, Morten; Hansen, Lars Kai.

In: NeuroImage, Vol. 100, 2014, p. 301-315.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Non-parametric Bayesian graph models reveal community structure in resting state fMRI

AU - Andersen, Kasper Winther

AU - Madsen, Kristoffer H.

AU - Siebner, Hartwig Roman

AU - Schmidt, Mikkel Nørgaard

AU - Mørup, Morten

AU - Hansen, Lars Kai

PY - 2014

Y1 - 2014

N2 - Modeling of resting state functional magnetic resonance imaging (rs-fMRI) data using network models is of increasing interest. It is often desirable to group nodes into clusters to interpret the communication patterns between nodes. In this study we consider three different nonparametric Bayesian models for node clustering in complex networks. In particular, we test their ability to predict unseen data and their ability to reproduce clustering across datasets. The three generative models considered are the Infinite Relational Model (IRM), Bayesian Community Detection (BCD), and the Infinite Diagonal Model (IDM). The models define probabilities of generating links within and between clusters and the difference between the models lies in the restrictions they impose upon the between-cluster link probabilities. IRM is the most flexible model with no restrictions on the probabilities of links between clusters. BCD restricts the between-cluster link probabilities to be strictly lower than within-cluster link probabilities to conform to the community structure typically seen in social networks. IDM only models a single between-cluster link probability, which can be interpreted as a background noise probability. These probabilistic models are compared against three other approaches for node clustering, namely Infomap, Louvain modularity, and hierarchical clustering. Using 3 different datasets comprising healthy volunteers' rs-fMRI we found that the BCD model was in general the most predictive and reproducible model. This suggests that rs-fMRI data exhibits community structure and furthermore points to the significance of modeling heterogeneous between-cluster link probabilities.

AB - Modeling of resting state functional magnetic resonance imaging (rs-fMRI) data using network models is of increasing interest. It is often desirable to group nodes into clusters to interpret the communication patterns between nodes. In this study we consider three different nonparametric Bayesian models for node clustering in complex networks. In particular, we test their ability to predict unseen data and their ability to reproduce clustering across datasets. The three generative models considered are the Infinite Relational Model (IRM), Bayesian Community Detection (BCD), and the Infinite Diagonal Model (IDM). The models define probabilities of generating links within and between clusters and the difference between the models lies in the restrictions they impose upon the between-cluster link probabilities. IRM is the most flexible model with no restrictions on the probabilities of links between clusters. BCD restricts the between-cluster link probabilities to be strictly lower than within-cluster link probabilities to conform to the community structure typically seen in social networks. IDM only models a single between-cluster link probability, which can be interpreted as a background noise probability. These probabilistic models are compared against three other approaches for node clustering, namely Infomap, Louvain modularity, and hierarchical clustering. Using 3 different datasets comprising healthy volunteers' rs-fMRI we found that the BCD model was in general the most predictive and reproducible model. This suggests that rs-fMRI data exhibits community structure and furthermore points to the significance of modeling heterogeneous between-cluster link probabilities.

KW - Complex network

KW - Graph theory

KW - Infinite Relational Model

KW - Bayesian Community Detection

KW - Resting state fMRI

U2 - 10.1016/j.neuroimage.2014.05.083

DO - 10.1016/j.neuroimage.2014.05.083

M3 - Journal article

VL - 100

SP - 301

EP - 315

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

ER -