@article {Gilson531830, author = {M Gilson and G Zamora-L{\'o}pez and V Pallar{\'e}s and MH Adhikari and M Senden and A Tauste Campo and D Mantini and M Corbetta and G Deco and A Insabato}, title = {MOU-EC: model-based whole-brain effective connectivity to extract biomarkers for brain dynamics from fMRI data and study distributed cognition}, elocation-id = {531830}, year = {2019}, doi = {10.1101/531830}, publisher = {Cold Spring Harbor Laboratory}, abstract = {Neuroimaging techniques are increasingly used to study brain cognition in humans. Beyond their individual activation, the functional associations between brain areas have become a standard proxy to describe how information is distributed across the brain network. Among the many analysis tools available, dynamic models of brain activity have been developed to overcome the limitations of original connectivity measures such as functional connectivity. In particular, much effort has been devoted to the assessment of directional interactions between brain areas from their observed activity. This paper summarizes our recent approach to analyze fMRI data based on our whole-brain effective connectivity referred to as MOU-EC, while discussing the pros and cons of its underlying assumptions with respect to other established approaches. Once tuned, the model provides a connectivity measure that reflects the dynamical state of BOLD activity obtained using fMRI, which can be used to explore the brain cognition. We focus on two important applications. First, as a connectivity measure, MOU-EC can be used to extract biomarkers for task-specific brain coordination, understood as the patterns of areas exchanging information. The multivariate nature of connectivity measures raises several challenges for whole-brain analysis, for which machine-learning tools presents some advantages over statistical testing. Second, we show how to interpret changes in MOU-EC connections in a collective and model-based manner, bridging with network analysis. To illustrate our framework, we use a dataset where subjects were recorded in two conditions, watching a movie and a black screen (referred to as rest). Our framework provides a comprehensive set of tools that open exciting perspectives for the study of distributed cognition, as well as neuropathologies.Abbreviations:Generative modelModel of (dynamic) equations that generates a signal to be fit to empirical data. This definition is different from the definition in statistics where a generative model describes the joint probability distribution of observed and predicted variables (here the network activity), as opposed to a discriminative model that describes the conditional probability of the predicted variables with respect to the observed variables.ObservableMeasure of model activity or applied to empirical data.Graphical modelGenerative model of Gaussian variables with linear interactions. Its output variables have a flat autocovariance (apart from zero time lag) and is used to model noisy data without temporal structure.Multivariate Ornstein-Uhlenbeck (MOU) processDynamic generative model in continuous time with linear interactions. Its output variables have both spatial and temporal correlations.Multivariate autoregressive (MAR) processDynamic generative model in discrete time with linear interactions. It is the equivalent of the MOU process in discrete time.Lyapunov optimization or natural gradient descentTuning procedure for the EC weights in the MOU network that fits the model FC covariance matrices to their empirical counterparts.Classification pipelineSuccession of machine-learning algorithms that aims to learn the mapping from input vectors to output labels (or categories). Here we use neuroimaging connectivity measures to predict cognitive conditions (like the task performed by a subject).Multinomial logistic regression (MLR)Linear classifier that can be efficiently trained using a gradient descent.k-nearest neighbour (kNN)Classifier that uses k known specimens per category to predict the category of new samples based on a similarity measure.BiomarkerSubset of observed features (here EC/FC links) that enable a robust classification.CommunicabilityMeasure of pairwise interactions between nodes (ROIs) in a network that takes indirect paths into account. In the present study, it corresponds to interactions over time for the MOU model.FlowExtension of communicability that incorporates the effect of input properties in the MOU model.}, URL = {https://www.biorxiv.org/content/early/2019/01/27/531830}, eprint = {https://www.biorxiv.org/content/early/2019/01/27/531830.full.pdf}, journal = {bioRxiv} }