Bayesian switching factor analysis for estimating time-varying functional connectivity in fMRI

ABSTRACT There is growing interest in understanding the dynamical properties of functional interactions between distributed brain regions. However, robust estimation of temporal dynamics from functional magnetic resonance imaging (fMRI) data remains challenging due to limitations in extant multivariate methods for modeling time‐varying functional interactions between multiple brain areas. Here, we develop a Bayesian generative model for fMRI time‐series within the framework of hidden Markov models (HMMs). The model is a dynamic variant of the static factor analysis model (Ghahramani and Beal, 2000). We refer to this model as Bayesian switching factor analysis (BSFA) as it integrates factor analysis into a generative HMM in a unified Bayesian framework. In BSFA, brain dynamic functional networks are represented by latent states which are learnt from the data. Crucially, BSFA is a generative model which estimates the temporal evolution of brain states and transition probabilities between states as a function of time. An attractive feature of BSFA is the automatic determination of the number of latent states via Bayesian model selection arising from penalization of excessively complex models. Key features of BSFA are validated using extensive simulations on carefully designed synthetic data. We further validate BSFA using fingerprint analysis of multisession resting‐state fMRI data from the Human Connectome Project (HCP). Our results show that modeling temporal dependencies in the generative model of BSFA results in improved fingerprinting of individual participants. Finally, we apply BSFA to elucidate the dynamic functional organization of the salience, central‐executive, and default mode networks—three core neurocognitive systems with central role in cognitive and affective information processing (Menon, 2011). Across two HCP sessions, we demonstrate a high level of dynamic interactions between these networks and determine that the salience network has the highest temporal flexibility among the three networks. Our proposed methods provide a novel and powerful generative model for investigating dynamic brain connectivity.

[1]  V. Menon.  Large-scale brain networks and psychopathology: a unifying triple network model , 2011, Trends in Cognitive Sciences.

[2]  David J. C. MacKay,et al.  Bayesian Methods for Backpropagation Networks , 1996 .

[3]  B. Biswal,et al.  Functional connectivity in the motor cortex of resting human brain using echo‐planar mri , 1995, Magnetic resonance in medicine.

[4]  Zoubin Ghahramani,et al.  Variational Inference for Bayesian Mixtures of Factor Analysers , 1999, NIPS.

[5]  Kaustubh Supekar,et al.  Distinct Global Brain Dynamics and Spatiotemporal Organization of the Salience Network , 2016, PLoS biology.

[6]  M. Chun,et al.  A neuromarker of sustained attention from whole-brain functional connectivity , 2015, Nature Neuroscience.

[7]  Michael I. Jordan,et al.  Graphical Models, Exponential Families, and Variational Inference , 2008, Found. Trends Mach. Learn..

[8]  M. Chun,et al.  Functional connectome fingerprinting: Identifying individuals based on patterns of brain connectivity , 2015, Nature Neuroscience.

[9]  M. Greicius,et al.  Decoding subject-driven cognitive states with whole-brain connectivity patterns. , 2012, Cerebral cortex.

[10]  Tor D. Wager,et al.  Dynamic functional connectivity using state-based dynamic community structure: Method and application to opioid analgesia , 2015, NeuroImage.

[11]  M. Chun,et al.  Functional connectome fingerprinting: Identifying individuals based on patterns of brain connectivity , 2015, Nature Neuroscience.

[12]  Mark W. Woolrich,et al.  Spectrally resolved fast transient brain states in electrophysiological data , 2016, NeuroImage.

[13]  Fenna M. Krienen,et al.  Opportunities and limitations of intrinsic functional connectivity MRI , 2013, Nature Neuroscience.

[14]  Eswar Damaraju,et al.  Tracking whole-brain connectivity dynamics in the resting state. , 2014, Cerebral cortex.

[15]  Geoffrey J. Gordon,et al.  A Unified View of Matrix Factorization Models , 2008, ECML/PKDD.

[16]  John C Gore,et al.  Functional MRI and multivariate autoregressive models. , 2010, Magnetic resonance imaging.

[17]  Jonathan D. Power,et al.  Intrinsic and Task-Evoked Network Architectures of the Human Brain , 2014, Neuron.

[18]  Karl J. Friston,et al.  Multivariate Autoregressive Modelling of fMRI time series , 2003 .

[19]  Lawrence R. Rabiner,et al.  A tutorial on hidden Markov models and selected applications in speech recognition , 1989, Proc. IEEE.

[20]  Dinggang Shen,et al.  A Hybrid of Deep Network and Hidden Markov Model for MCI Identification with Resting-State fMRI , 2015, MICCAI.

[21]  Karl J. Friston,et al.  Principles of Brain Dynamics: Global State Interactions , 2012 .

[22]  Geoffrey E. Hinton,et al.  Bayesian Learning for Neural Networks , 1995 .

[23]  S. Petersen,et al.  A dual-networks architecture of top-down control , 2008, Trends in Cognitive Sciences.

[24]  Dimitri Van De Ville,et al.  On spurious and real fluctuations of dynamic functional connectivity during rest , 2015, NeuroImage.

[25]  Jean-Loup Guillaume,et al.  Fast unfolding of communities in large networks , 2008, 0803.0476.

[26]  Matthew J. Beal.  Variational algorithms for approximate Bayesian inference , 2003 .

[27]  C. Li,et al.  Causal Interactions Within a Frontal-Cingulate-Parietal Network During Cognitive Control: Convergent Evidence from a Multisite-Multitask Investigation. , 2016, Cerebral cortex.

[28]  Michael I. Jordan,et al.  An Introduction to Variational Methods for Graphical Models , 1999, Machine-mediated learning.

[29]  Kevin P. Murphy,et al.  Machine learning - a probabilistic perspective , 2012, Adaptive computation and machine learning series.

[30]  Martin A. Lindquist,et al.  Evaluating dynamic bivariate correlations in resting-state fMRI: A comparison study and a new approach , 2014, NeuroImage.

[31]  Carl E. Rasmussen,et al.  Bayesian Modelling of fMRI lime Series , 1999, NIPS.

[32]  Martin A. Lindquist,et al.  Dynamic connectivity regression: Determining state-related changes in brain connectivity , 2012, NeuroImage.

[33]  Jonathan D. Power,et al.  Evidence for Hubs in Human Functional Brain Networks , 2013, Neuron.

[34]  Christos Davatzikos,et al.  Unsupervised Learning of Functional Network Dynamics in Resting State fMRI , 2013, IPMI.

[35]  S. Balqis Samdin,et al.  A Unified Estimation Framework for State-Related Changes in Effective Brain Connectivity , 2017, IEEE Transactions on Biomedical Engineering.

[36]  Vinod Menon,et al.  Large-Scale Functional Brain Organization , 2015 .

[37]  Stephen M. Smith,et al.  Temporally-independent functional modes of spontaneous brain activity , 2012, Proceedings of the National Academy of Sciences.

[38]  Radford M. Neal.  Pattern Recognition and Machine Learning , 2007, Technometrics.

[39]  Hernando Ombao,et al.  Modeling the Evolution of Dynamic Brain Processes During an Associative Learning Experiment , 2016 .

[40]  Karl J. Friston,et al.  The Dynamic Brain: From Spiking Neurons to Neural Masses and Cortical Fields , 2008, PLoS Comput. Biol..

[41]  Eric Postma,et al.  Combining Structural Connectivity and Response Latencies to Model the Structure of the Visual System , 2008, PLoS Comput. Biol..

[42]  V. Calhoun,et al.  The Chronnectome: Time-Varying Connectivity Networks as the Next Frontier in fMRI Data Discovery , 2014, Neuron.

[43]  Kewei Chen,et al.  Identification and validation of effective connectivity networks in functional magnetic resonance imaging using switching linear dynamic systems , 2010, NeuroImage.

[44]  Vinod Menon,et al.  Functional connectivity in the resting brain: A network analysis of the default mode hypothesis , 2002, Proceedings of the National Academy of Sciences of the United States of America.

[45]  Geoffrey E. Hinton,et al.  The EM algorithm for mixtures of factor analyzers , 1996 .

[46]  Stephen M Smith,et al.  Fast transient networks in spontaneous human brain activity , 2014, eLife.

[47]  Brian Everitt,et al.  An Introduction to Latent Variable Models , 1984 .

[48]  M. Willig,et al.  Standardized Assessment of Biodiversity Trends in Tropical Forest Protected Areas: The End Is Not in Sight , 2016, PLoS biology.

[49]  Tommi S. Jaakkola,et al.  Tutorial on variational approximation methods , 2000 .

[50]  M. Breakspear.  "Dynamic" connectivity in neural systems: theoretical and empirical considerations. , 2004, Neuroinformatics.

[51]  Dimitri Van De Ville,et al.  Transient brain activity disentangles fMRI resting-state dynamics in terms of spatially and temporally overlapping networks , 2015, Nature Communications.

[52]  Leonardo L. Gollo,et al.  Time-resolved resting-state brain networks , 2014, Proceedings of the National Academy of Sciences.

[53]  Kaustubh Supekar,et al.  Temporal Dynamics and Developmental Maturation of Salience, Default and Central-Executive Network Interactions Revealed by Variational Bayes Hidden Markov Modeling , 2016, PLoS Comput. Biol..

[54]  V. Menon,et al.  Saliency, switching, attention and control: a network model of insula function , 2010, Brain Structure and Function.

[55]  Chee-Ming Ting,et al.  Is First-Order Vector Autoregressive Model Optimal for fMRI Data? , 2015, Neural Computation.

[56]  Catie Chang,et al.  Time–frequency dynamics of resting-state brain connectivity measured with fMRI , 2010, NeuroImage.

[57]  G. Glover,et al.  Dissociable Intrinsic Connectivity Networks for Salience Processing and Executive Control , 2007, The Journal of Neuroscience.

[58]  ModelsMICHAEL,et al.  An Introduction to Variational Methods forGraphical , 1998 .