State-space model with deep learning for functional dynamics estimation in resting-state fMRI

Studies on resting-state functional Magnetic Resonance Imaging (rs-fMRI) have shown that different brain regions still actively interact with each other while a subject is at rest, and such functional interaction is not stationary but changes over time. In terms of a large-scale brain network, in this paper, we focus on time-varying patterns of functional networks, i.e., functional dynamics, inherent in rs-fMRI, which is one of the emerging issues along with the network modelling. Specifically, we propose a novel methodological architecture that combines deep learning and state-space modelling, and apply it to rs-fMRI based Mild Cognitive Impairment (MCI) diagnosis. We first devise a Deep Auto-Encoder (DAE) to discover hierarchical non-linear functional relations among regions, by which we transform the regional features into an embedding space, whose bases are complex functional networks. Given the embedded functional features, we then use a Hidden Markov Model (HMM) to estimate dynamic characteristics of functional networks inherent in rs-fMRI via internal states, which are unobservable but can be inferred from observations statistically. By building a generative model with an HMM, we estimate the likelihood of the input features of rs-fMRI as belonging to the corresponding status, i.e., MCI or normal healthy control, based on which we identify the clinical label of a testing subject. In order to validate the effectiveness of the proposed method, we performed experiments on two different datasets and compared with state-of-the-art methods in the literature. We also analyzed the functional networks learned by DAE, estimated the functional connectivities by decoding hidden states in HMM, and investigated the estimated functional connectivities by means of a graph-theoretic approach.

[1]  Archana Venkataraman,et al.  Intrinsic functional connectivity as a tool for human connectomics: theory, properties, and optimization. , 2010, Journal of neurophysiology.

[2]  Fuhui Long,et al.  Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy , 2003, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Sylvain Faisan,et al.  Hidden Markov multiple event sequence models: A paradigm for the spatio-temporal analysis of fMRI data , 2007, Medical Image Anal..

[4]  Daoqiang Zhang,et al.  Group-constrained sparse fMRI connectivity modeling for mild cognitive impairment identification , 2013, Brain Structure and Function.

[5]  Zhijun Zhang,et al.  Abnormal resting-state functional connectivity of posterior cingulate cortex in amnestic type mild cognitive impairment , 2009, Brain Research.

[6]  Yuan Zhou,et al.  Abnormal Cortical Networks in Mild Cognitive Impairment and Alzheimer's Disease , 2010, PLoS Comput. Biol..

[7]  Daniel L. Rubin,et al.  Network Analysis of Intrinsic Functional Brain Connectivity in Alzheimer's Disease , 2008, PLoS Comput. Biol..

[8]  W. Thies,et al.  2008 Alzheimer’s disease facts and figures , 2008, Alzheimer's & Dementia.

[9]  V. Calhoun,et al.  Aberrant "default mode" functional connectivity in schizophrenia. , 2007, The American journal of psychiatry.

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

[11]  Raghu Machiraju,et al.  Spatio-temporal models of mental processes from fMRI , 2011, NeuroImage.

[12]  M. Lowe,et al.  Activity and Connectivity of Brain Mood Regulating Circuit in Depression: A Functional Magnetic Resonance Study , 2005, Biological Psychiatry.

[13]  Scott T. Grafton,et al.  Dynamic reconfiguration of human brain networks during learning , 2010, Proceedings of the National Academy of Sciences.

[14]  Seong-Whan Lee,et al.  Hierarchical feature representation and multimodal fusion with deep learning for AD/MCI diagnosis , 2014, NeuroImage.

[15]  Olaf Sporns,et al.  Complex network measures of brain connectivity: Uses and interpretations , 2010, NeuroImage.

[16]  Vince D. Calhoun,et al.  Restricted Boltzmann machines for neuroimaging: An application in identifying intrinsic networks , 2014, NeuroImage.

[17]  Yong He,et al.  Effects of Different Correlation Metrics and Preprocessing Factors on Small-World Brain Functional Networks: A Resting-State Functional MRI Study , 2012, PloS one.

[18]  Yuan Zhou,et al.  Widespread functional disconnectivity in schizophrenia with resting-state functional magnetic resonance imaging , 2006, Neuroreport.

[19]  Kevin Murphy,et al.  The impact of global signal regression on resting state correlations: Are anti-correlated networks introduced? , 2009, NeuroImage.

[20]  Alex Chase Alzheimer disease: Altered functional connectivity in preclinical dementia , 2014, Nature Reviews Neurology.

[21]  Geoffrey E. Hinton Training Products of Experts by Minimizing Contrastive Divergence , 2002, Neural Computation.

[22]  V. Calhoun,et al.  Selective changes of resting-state networks in individuals at risk for Alzheimer's disease , 2007, Proceedings of the National Academy of Sciences.

[23]  E. Bullmore,et al.  Behavioral / Systems / Cognitive Functional Connectivity and Brain Networks in Schizophrenia , 2010 .

[24]  Geoffrey E. Hinton A Practical Guide to Training Restricted Boltzmann Machines , 2012, Neural Networks: Tricks of the Trade.

[25]  Vince D. Calhoun,et al.  Deep learning for neuroimaging: a validation study , 2013, Front. Neurosci..

[26]  S. Black,et al.  Evidence from Functional Neuroimaging of a Compensatory Prefrontal Network in Alzheimer's Disease , 2003, The Journal of Neuroscience.

[27]  C. Stam Modern network science of neurological disorders , 2014, Nature Reviews Neuroscience.

[28]  M. Greicius Resting-state functional connectivity in neuropsychiatric disorders , 2008, Current opinion in neurology.

[29]  John Pluta,et al.  Increased functional connectivity within medial temporal lobe in mild cognitive impairment , 2013, Hippocampus.

[30]  J. M. Moran,et al.  Local and long-range functional connectivity is reduced in concert in autism spectrum disorders , 2013, Proceedings of the National Academy of Sciences.

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

[32]  Yuan Zhou,et al.  Functional disintegration in paranoid schizophrenia using resting-state fMRI , 2007, Schizophrenia Research.

[33]  S. Debener,et al.  Default-mode brain dysfunction in mental disorders: A systematic review , 2009, Neuroscience & Biobehavioral Reviews.

[34]  M. Posner,et al.  The Development of Executive Attention: Contributions to the Emergence of Self-Regulation , 2005, Developmental neuropsychology.

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

[36]  Yee Whye Teh,et al.  A Fast Learning Algorithm for Deep Belief Nets , 2006, Neural Computation.

[37]  M. Greicius,et al.  Default-mode network activity distinguishes Alzheimer's disease from healthy aging: Evidence from functional MRI , 2004, Proc. Natl. Acad. Sci. USA.

[38]  G. Glover,et al.  Resting-State Functional Connectivity in Major Depression: Abnormally Increased Contributions from Subgenual Cingulate Cortex and Thalamus , 2007, Biological Psychiatry.

[39]  Gaohong Wu,et al.  Alzheimer Disease: evaluation of a functional MR imaging index as a marker. , 2002, Radiology.

[40]  D. Schacter,et al.  The Brain's Default Network , 2008, Annals of the New York Academy of Sciences.

[41]  D. Shen,et al.  Discriminant analysis of longitudinal cortical thickness changes in Alzheimer's disease using dynamic and network features , 2012, Neurobiology of Aging.

[42]  S. Rombouts,et al.  Altered resting state networks in mild cognitive impairment and mild Alzheimer's disease: An fMRI study , 2005, Human brain mapping.

[43]  Vince D. Calhoun,et al.  Deep neural network with weight sparsity control and pre-training extracts hierarchical features and enhances classification performance: Evidence from whole-brain resting-state functional connectivity patterns of schizophrenia , 2016, NeuroImage.

[44]  Seong-Whan Lee,et al.  Latent feature representation with stacked auto-encoder for AD/MCI diagnosis , 2013, Brain Structure and Function.

[45]  Indrayana Rustandi,et al.  Modeling fMRI data generated by overlapping cognitive processes with unknown onsets using Hidden Process Models , 2009, NeuroImage.

[46]  Vince D. Calhoun,et al.  Alterations in Memory Networks in Mild Cognitive Impairment and Alzheimer's Disease: An Independent Component Analysis , 2006, The Journal of Neuroscience.

[47]  Karl J. Friston,et al.  Functional Connectivity: The Principal-Component Analysis of Large (PET) Data Sets , 1993, Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism.

[48]  Maurizio Corbetta,et al.  The human brain is intrinsically organized into dynamic, anticorrelated functional networks. , 2005, Proceedings of the National Academy of Sciences of the United States of America.

[49]  David A. Leopold,et al.  Dynamic functional connectivity: Promise, issues, and interpretations , 2013, NeuroImage.

[50]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[51]  M. Fox,et al.  The global signal and observed anticorrelated resting state brain networks. , 2009, Journal of neurophysiology.

[52]  J. Whitwell,et al.  Alzheimer's disease neuroimaging , 2018, Current opinion in neurology.

[53]  Kuncheng Li,et al.  Altered functional connectivity in early Alzheimer's disease: A resting‐state fMRI study , 2007, Human brain mapping.

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

[55]  Scott Peltier,et al.  Abnormalities of intrinsic functional connectivity in autism spectrum disorders, , 2009, NeuroImage.

[56]  Dimitri Van De Ville,et al.  Principal components of functional connectivity: A new approach to study dynamic brain connectivity during rest , 2013, NeuroImage.

[57]  R Cameron Craddock,et al.  Disease state prediction from resting state functional connectivity , 2009, Magnetic resonance in medicine.

[58]  M. Koskinen,et al.  Aberrant temporal and spatial brain activity during rest in patients with chronic pain , 2010, Proceedings of the National Academy of Sciences.

[59]  Bill Wilson,et al.  Intensity of dementia through latent variable modelling (I-DeLV) in the AIBL cohort , 2012, Alzheimer's & Dementia.

[60]  Dinggang Shen,et al.  Supervised Discriminative Group Sparse Representation for Mild Cognitive Impairment Diagnosis , 2014, Neuroinformatics.

[61]  Daoqiang Zhang,et al.  Predicting Future Clinical Changes of MCI Patients Using Longitudinal and Multimodal Biomarkers , 2012, PloS one.

[62]  Geoffrey E. Hinton,et al.  Reducing the Dimensionality of Data with Neural Networks , 2006, Science.

[63]  Vince D. Calhoun,et al.  A method for functional network connectivity among spatially independent resting-state components in schizophrenia , 2008, NeuroImage.

[64]  N. Tzourio-Mazoyer,et al.  Automated Anatomical Labeling of Activations in SPM Using a Macroscopic Anatomical Parcellation of the MNI MRI Single-Subject Brain , 2002, NeuroImage.

[65]  Daniel A. Handwerker,et al.  Periodic changes in fMRI connectivity , 2012, NeuroImage.

[66]  Yong He,et al.  BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics , 2013, PloS one.

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

[68]  Dinggang Shen,et al.  Hierarchical multi-atlas label fusion with multi-scale feature representation and label-specific patch partition , 2015, NeuroImage.