Bayesian structure learning for dynamic brain connectivity

Michael Riis Andersen, Ole Winther, Lars Kai Hansen, Russell Poldrack, Oluwasanmi Oluseye Koyejo

Research output: Contribution to conferencePaper

Abstract

Human brain activity as measured by fMRI exhibits strong correlations between brain regions which are believed to vary over time. Importantly, dynamic connectivity has been linked to individual differences in physiology, psychology and behavior, and has shown promise as a biomarker for disease. The state of the art in computational neuroimaging is to estimate the brain networks as relatively short sliding window covariance matrices, which leads to high variance estimates, thereby resulting in high overall error. This manuscript proposes a novel Bayesian model for dynamic brain connectivity. Motivated by the underlying neuroscience, the model estimates covariances which vary smoothly over time, with an instantaneous decomposition into a collection of spatially sparse components – resulting in parsimonious and highly interpretable estimates of dynamic brain connectivity. Simulated results are presented to illustrate the performance of the model even when it is mis-specified. For real brain imaging data with unknown ground truth, in addition to qualitative evaluation, we devise a simple classification task which suggests that the estimated brain networks better capture the underlying structure.

Original languageEnglish (US)
Pages1436-1446
Number of pages11
StatePublished - Jan 1 2018
Event21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018 - Playa Blanca, Lanzarote, Canary Islands, Spain
Duration: Apr 9 2018Apr 11 2018

Conference

Conference21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018
CountrySpain
CityPlaya Blanca, Lanzarote, Canary Islands
Period4/9/184/11/18

Fingerprint

Bayesian Learning
Structure Learning
Brain
Connectivity
Estimate
Vary
Neuroimaging
Individual Differences
Functional Magnetic Resonance Imaging
Neuroscience
Sliding Window
Physiology
Biomarkers
Bayesian Model
Covariance matrix
Instantaneous
Imaging
Decomposition
Imaging techniques
Decompose

ASJC Scopus subject areas

  • Statistics and Probability
  • Artificial Intelligence

Cite this

Andersen, M. R., Winther, O., Hansen, L. K., Poldrack, R., & Koyejo, O. O. (2018). Bayesian structure learning for dynamic brain connectivity. 1436-1446. Paper presented at 21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018, Playa Blanca, Lanzarote, Canary Islands, Spain.

Bayesian structure learning for dynamic brain connectivity. / Andersen, Michael Riis; Winther, Ole; Hansen, Lars Kai; Poldrack, Russell; Koyejo, Oluwasanmi Oluseye.

2018. 1436-1446 Paper presented at 21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018, Playa Blanca, Lanzarote, Canary Islands, Spain.

Research output: Contribution to conferencePaper

Andersen, MR, Winther, O, Hansen, LK, Poldrack, R & Koyejo, OO 2018, 'Bayesian structure learning for dynamic brain connectivity' Paper presented at 21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018, Playa Blanca, Lanzarote, Canary Islands, Spain, 4/9/18 - 4/11/18, pp. 1436-1446.
Andersen MR, Winther O, Hansen LK, Poldrack R, Koyejo OO. Bayesian structure learning for dynamic brain connectivity. 2018. Paper presented at 21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018, Playa Blanca, Lanzarote, Canary Islands, Spain.
Andersen, Michael Riis ; Winther, Ole ; Hansen, Lars Kai ; Poldrack, Russell ; Koyejo, Oluwasanmi Oluseye. / Bayesian structure learning for dynamic brain connectivity. Paper presented at 21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018, Playa Blanca, Lanzarote, Canary Islands, Spain.11 p.
@conference{d1dfc38964ab4d9e8c8aa2063e5f7e84,
title = "Bayesian structure learning for dynamic brain connectivity",
abstract = "Human brain activity as measured by fMRI exhibits strong correlations between brain regions which are believed to vary over time. Importantly, dynamic connectivity has been linked to individual differences in physiology, psychology and behavior, and has shown promise as a biomarker for disease. The state of the art in computational neuroimaging is to estimate the brain networks as relatively short sliding window covariance matrices, which leads to high variance estimates, thereby resulting in high overall error. This manuscript proposes a novel Bayesian model for dynamic brain connectivity. Motivated by the underlying neuroscience, the model estimates covariances which vary smoothly over time, with an instantaneous decomposition into a collection of spatially sparse components – resulting in parsimonious and highly interpretable estimates of dynamic brain connectivity. Simulated results are presented to illustrate the performance of the model even when it is mis-specified. For real brain imaging data with unknown ground truth, in addition to qualitative evaluation, we devise a simple classification task which suggests that the estimated brain networks better capture the underlying structure.",
author = "Andersen, {Michael Riis} and Ole Winther and Hansen, {Lars Kai} and Russell Poldrack and Koyejo, {Oluwasanmi Oluseye}",
year = "2018",
month = "1",
day = "1",
language = "English (US)",
pages = "1436--1446",
note = "21st International Conference on Artificial Intelligence and Statistics, AISTATS 2018 ; Conference date: 09-04-2018 Through 11-04-2018",

}

TY - CONF

T1 - Bayesian structure learning for dynamic brain connectivity

AU - Andersen, Michael Riis

AU - Winther, Ole

AU - Hansen, Lars Kai

AU - Poldrack, Russell

AU - Koyejo, Oluwasanmi Oluseye

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Human brain activity as measured by fMRI exhibits strong correlations between brain regions which are believed to vary over time. Importantly, dynamic connectivity has been linked to individual differences in physiology, psychology and behavior, and has shown promise as a biomarker for disease. The state of the art in computational neuroimaging is to estimate the brain networks as relatively short sliding window covariance matrices, which leads to high variance estimates, thereby resulting in high overall error. This manuscript proposes a novel Bayesian model for dynamic brain connectivity. Motivated by the underlying neuroscience, the model estimates covariances which vary smoothly over time, with an instantaneous decomposition into a collection of spatially sparse components – resulting in parsimonious and highly interpretable estimates of dynamic brain connectivity. Simulated results are presented to illustrate the performance of the model even when it is mis-specified. For real brain imaging data with unknown ground truth, in addition to qualitative evaluation, we devise a simple classification task which suggests that the estimated brain networks better capture the underlying structure.

AB - Human brain activity as measured by fMRI exhibits strong correlations between brain regions which are believed to vary over time. Importantly, dynamic connectivity has been linked to individual differences in physiology, psychology and behavior, and has shown promise as a biomarker for disease. The state of the art in computational neuroimaging is to estimate the brain networks as relatively short sliding window covariance matrices, which leads to high variance estimates, thereby resulting in high overall error. This manuscript proposes a novel Bayesian model for dynamic brain connectivity. Motivated by the underlying neuroscience, the model estimates covariances which vary smoothly over time, with an instantaneous decomposition into a collection of spatially sparse components – resulting in parsimonious and highly interpretable estimates of dynamic brain connectivity. Simulated results are presented to illustrate the performance of the model even when it is mis-specified. For real brain imaging data with unknown ground truth, in addition to qualitative evaluation, we devise a simple classification task which suggests that the estimated brain networks better capture the underlying structure.

UR - http://www.scopus.com/inward/record.url?scp=85067805909&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85067805909&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85067805909

SP - 1436

EP - 1446

ER -