Dynamic causal modeling

This sandbox is in the article namespace. Either move this page into your userspace, or remove the {{User sandbox}} template. Dynamic Causal Modelling (DCM) is a methodology and software framework for specifying models of neural dynamics, estimating their parameters and comparing their evidence ^[1]. It enables hypotheses to be tested about the interaction of neural populations (effective connectivity) using functional neuroimaging data e.g., functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG), electroencephalography (EEG) or intracranial Local Field Potentials (LFP).

Procedure

DCM is used to estimate the coupling among brain regions and the changes in coupling due to experimental changes (e.g., time or context). The basic idea is to construct reasonably realistic models of interacting brain regions. These models are then supplemented with a forward model of how the hidden states of each brain region (e.g., neuronal activity) cause the measured responses. This enables the best model(s) and their parameters (i.e., effective connectivity) to be identified from observed data. Bayesian model comparison is used to compare models based on their their evidence, which can then be characterised in terms of their parameters (e.g. connection strengths).

Experiments using DCM typically involve the following stages ^[2]:

Experimental design. Formulate specific hypotheses and conduct a neuroimaging experiment.
Data preparation. Pre-process the acquired data (e.g. select relevant data features and remove confounds).
Model specification. Specify one or more forward models (DCMs) of how the data were caused.
Model estimation. Fit the model(s) to the data to determine their evidence and parameters.
Model comparison. Compare the evidence for the models using Bayesian Model Comparison, at the single-subject level or at the group level, and inspect the parameters of the model(s).

The key steps are briefly reviewed below.

Experimental design

Functional neuroimaging experiments are typically task-based or resting state. In task-based experiments, brain responses are evoked by known deterministic inputs (experimentally controlled stimuli) that embody designed changes in sensory stimulation or cognitive set. These experimental or exogenous variables can change neural activity in one of two ways. First, they can elicit responses through direct influences on specific brain regions. This would include, for example, sensory evoked responses in the early visual cortex. The second class of inputs exerts their effects vicariously, through a modulation of the coupling among nodes, for example, the influence of attention on the processing of sensory information. These two types of input - driving and modulatory - are separately parameterized in DCM. To enable efficient estimation of driving and modulatory effects, a 2x2 factorial experimental design is often used - with one factor modelled as the driving input and the other as the modulatory input ^[2].

Resting state experiments have no experimental manipulations within the period of the neuroimaging recording. Instead, endogenous fluctuations in brain connectivity during the scan are of interest, or the differences in connectivity between scans or subjects. The DCM framework includes models and procedures for resting state data, described below..

Model specification

Dynamic Causal Models (DCMs) are nonlinear state-space models in continuous time, parameterized in terms of directed effective connectivity between brain regions. Unlike Bayesian Networks, DCMs can be cyclic, and unlike Structural Equation modelling and Granger causality, DCM does not depend on the theory of Martingales, i.e., it does not assume that random fluctuations' are serially uncorrelated. Various models have been developed for use with DCM and the experimenter selects their preferred model based on the types of hypothesis they wish to address and the type of data they have collected.

Functional MRI

The neural model in DCM for fMRI uses a simple mathematical device - a Taylor approximation - which captures the gross causal causal influences between brain regions and their change due to driving and modulatory inputs. This is coupled with a detailed biophysical model of the generation of the BOLD response and the MRI signal, based on the Balloon model of Buxton et al.^[3] and extended for use in DCM ^[4]^[5]. The neural model was subsequently extended to include the interaction of excitatory and inhibitory neural populations ^[6] and non-linear influences of neural populations on the coupling between other populations^[7].

Support for resting state analysis was first introduced in Stochastic DCM^[8], which estimates both neural fluctuations and connectivity parameters in the time ___domain using a procedure called Generalized Filtering. A faster and more accurate solution was introduced which operates in the frequency ___domain, called DCM for Cross-Spectral Densities (CSD) ^[9]^[10]. Both of these can be applied to large-scale brain networks by using priors based on functional connectivity^[11]. Another recent development is Regression DCM. This also operates in the frequency ___domain, but linearizes the model under certain simplifications, such as a fixed (canonical) haemodynamic response function. The means that the model can be inverted rapidly as a General Linear Model and can be applied to large-scale brain networks.

EEG / MEG

EEG and MEG data can support the estimation of more biologically detailed neural models than fMRI, as they have richer dynamics with higher temporal resolution.

The predominant model is DCM for evoked responses (DCM for ERP). It is a biologically plausible neural mass model, based on earlier work by Jansen and Rit (1995) and Lopes da Silva et al. (1974), which emulates the activity of a cortical area using three neuronal subpopulations. Each subpopulation rests on two operators. The first transforms the pre-synaptic firing rate into a Post-Synaptic Potential (PSP), by convolving a synaptic response function (kernel) by the pre-input. As a result, this is referred to as a convolution model. The second operator, a sigmoid function, transforms the membrane potential into a firing rate of action potentials. DCM for LFP (Local Field Potentials) extended this model to include the effects of specific ion channels on spike generation.

A short paragraph on the CMC model please?

A short paragraph on conductance-based models please? A similar length to the one above. See Moran, Pinotsis and Friston (2013), Frontiers in Computational Neuroscience.

Model estimation

Model inversion or estimation is implemented in DCM using a variational Bayesian optimisation scheme. It provides two useful quantities. The log marginal likelihood or model evidence $\ln {p(y|m)}$ is the probability of observing of the given data under the model. This cannot be calculated exactly and in DCM it is approximated by a quantity called the negative variational free energy $F$ , referred to in machine learning as the Evidence Lower Bound (ELBO). Hypotheses are tested by comparing the evidence for different models based on their free energy, a procedure named Bayesian model comparison. Model estimation also provides estimates of the parameters $p(\theta |y)$ , for example the connection strengths, which maximise the free energy. Where models differ only in their priors, Bayesian Model Reduction can be used to rapidly the derive the evidence and parameters for nested or reduced models from a full model.

Model comparison

Neuroimaging studies typically investigate effects which are conserved at the group level, or which differ between subjects. There are two predominant approaches for group-level analysis: random effects Bayesian Model Selection (BMS) and Parametric Empirical Bayes (PEB). Random effects BMS posits that subjects differ in terms of which model generated their data - e.g. drawing a random subject from the population, there would be a 25% chance their data were generated by model 1 and a 75% chance their data were generated by model 2. The analysis pipeline for the BMS approach procedure follows a series of steps:

Specify and estimate multiple DCMs per subject, where each DCM (or set of DCMs) embodies a hypothesis.
Perform random effects BMS to estimate the proportion of subjects whose data were generated by each model
Perform Bayesian Model Averaging, which is a weighted average over the parameters of the DCMs. This means that models greater probability contribute more to the average than those with lower probability.

The most recently developed PEB approach is a hierarchical model over parameters (connection strengths). It eschews the notion of different models at the level of individual subjects, and posits that people differ in the (continuous) strength of their individual connections. The PEB approach separates sources of variability in connection strengths across subjects into hypothesised covariates and uninteresting between-subject variability (random effects). The PEB procedure is as follows:

Specify a single 'full' DCM per subject which contains all connectivity parameters of interest.
Model the estimated parameters (the full posterior density) from all subjects using a Bayesian General Linear Model at the group level.
Test hypotheses by comparing the full group-level model to reduced group-level models where certain combinations of connections have been switched off.

Limitations / drawbacks

DCM is a hypothesis-driven approach for investigating the interactions among pre-defined regions of interest. It is not ideally suited for exploratory analyses. Although methods have been implemented for automatic search over reduced models and for modelling large-scale brain networks, these methods expect clear hypotheses. Other approaches such as psycho-physical interactions (PPI) analysis may be more appropriate in contexts with less strong hypotheses.

The variational Bayesian methods used for model estimation used approximations based on the Laplace approximation that the parameters are normally distributed. This approximation can break down in the context of highly non-linear models, such as those used in EEG / MEG analysis, where local minima can preclude the free energy from serving as a close lower bound on log model evidence. The approximations used in this scheme can be evaluated using sampling approaches.

Software implementations

DCM is implemented in the Statistical Parametric Mapping software package, where it serves as the canonical or reference implementation (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/). It has been re-implemented and developed in the Tapas software collection (https://www.tnu.ethz.ch/en/software/tapas.html) and the VBA toolbox (http://mbb-team.github.io/VBA-toolbox/).

References

Acs, F., Greenlee, M.W., 2008. Connectivity modulation of early visual processing areas during covert and overt tracking tasks. Neuroimage 41, 380-388.

Akaike, H., 1985. Prediction and Entropy. In A. C. Atkinson and S. E. Feinberg (eds.), A Celebration of Statistics. New York: Springer. 1-24.

Allen, P., Mechelli, A., Stephan, K.E., Day, F., Dalton, J., Williams, S., McGuire, P.K., 2008. Fronto-temporal interactions during overt verbal initiation and suppression. J Cogn Neurosci 20, 1656-1669.

Box, G.E.P., 1980. ‘Sampling and Bayes’ Inference in Scientific Modelling and Robustness, J. Roy. Stat. Soc., Series A, Vol.143, 383-430.

Büchel, C., Friston, K.J., 1997. Modulation of connectivity in visual pathways by attention: cortical interactions evaluated with structural equation modelling and fMRI. Cereb Cortex 7, 768-778.

Buxton, R.B., Wong, E.C., Frank, L.R., 1998. Dynamics of blood flow and oxygenation changes during brain activation: the balloon model. Magn. Reson. Med. 39, 855-864.

Chen, C.C., Kiebel, S.J., Friston, K.J., 2008. Dynamic causal modelling of induced responses. Neuroimage 41, 1293-1312.

Daunizeau, J., Friston, K.J., 2007. A mesostate-space model for EEG and MEG. Neuroimage 38:67–81.

Daunizeau, J., Friston, K.J., Kiebel, S.J., 2009a. Variational Bayesian identification and prediction of stochastic nonlinear dynamic causal models. Physica, D 238, 2089–2118.

Daunizeau, J., Kiebel, S.J., Friston, K.J., 2009b. Dynamic causal modelling of distributed electromagnetic responses. Neuroimage 47, 590-601.

Daunizeau, J., David, O., Stephan, K.E., 2010. Dynamic Causal Modelling: a critical review of the biophysical and statistical foundations. Neuroimage, in press.

David, O., Friston, K.J., 2003. A neural mass model for MEG/EEG: coupling and neuronal dynamics. Neuroimage 20, 1743-1755.

David, O., Harrison, L., Friston, K.J., 2005. Modelling event-related responses in the brain. Neuroimage 25, 756-770.

David, O., Kiebel, S.J., Harrison, L.M., Mattout, J., Kilner, J.M., Friston, K.J., 2006. Dynamic causal modeling of evoked responses in EEG and MEG. Neuroimage 30, 1255-1272.

Dempster, A.P., Laird, N.M., Rubin, D.B., 1977. Maximum likelihood from incomplete data via EM algorithm. Journal of the Royal Statistical Society Series B-Methodological 39, 1-38.

Doeller, C.F., Opitz, B., Mecklinger, A., Krick, C., Reith, W., Schroger, E., 2003. Prefrontal cortex involvement in preattentive auditory deviance detection: neuroimaging and electrophysiological evidence. Neuroimage 20, 1270-1282.

Ethofer, T., Anders, S., Erb, M., Herbert, C., Wiethoff, S., Kissler, J., Grodd, W., Wildgruber, D., 2006. Cerebral pathways in processing of affective prosody: a dynamic causal modeling study. Neuroimage 30, 580-587.

Fairhall, S.L., Ishai, A., 2007. Effective connectivity within the distributed cortical network for face perception. Cereb Cortex 17, 2400-2406.

Felleman, D.J., Van Essen, D.C., 1991. Distributed hierarchical processing in the primate cerebral cortex. Cereb Cortex 1, 1-47.

Friston, K.J., Mechelli, A., Turner, R., Price, C.J., 2000. Nonlinear responses in fMRI: the Balloon model, Volterra kernels, and other hemodynamics. Neuroimage 12, 466-477.

Friston, K.J., 2002. Bayesian estimation of dynamical systems: an application to fMRI. Neuroimage 16, 513-530.

Friston, K.J., Harrison, L., Penny, W., 2003. Dynamic causal modelling. Neuroimage 19, 1273-1302.

Friston, K., Mattout, J., Trujillo-Barreto, N., Ashburner, J., Penny, W., 2007. Variational free energy and the Laplace approximation. Neuroimage 34, 220-234.

Friston, K.J., Trujillo-Barreto, N., Daunizeau, J., 2008. DEM: a variational treatment of dynamic systems. Neuroimage 41(3):849-85.

Friston, K., 2009. Causal modelling and brain connectivity in functional magnetic resonance imaging. PLoS Biol 7, e33.

Garrido, M.I., Kilner, J.M., Kiebel, S.J., Stephan, K.E., Friston, K.J., 2007. Dynamic causal modelling of evoked potentials: a reproducibility study. Neuroimage 36, 571-580.

Garrido, M.I., Friston, K.J., Kiebel, S.J., Stephan, K.E., Baldeweg, T., Kilner, J.M., 2008. The functional anatomy of the MMN: a DCM study of the roving paradigm. Neuroimage 42, 936-944.

Good, I.J., 1965. “The Estimation of Probabilities: An Essay on Modern Bayesian Methods”, Cambridge, Mass, MIT Press.

Grol, M.J., Majdandzic, J., Stephan, K.E., Verhagen, L., Dijkerman, H.C., Bekkering, H., Verstraten, F.A., Toni, I., 2007. Parieto-frontal connectivity during visually guided grasping. J Neurosci 27, 11877-11887.

Heim, S., Eickhoff, S.B., Ischebeck, A.K., Friederici, A.D., Stephan, K.E., Amunts, K., 2009. Effective connectivity of the left BA 44, BA 45, and inferior temporal gyrus during lexical and phonological decisions identified with DCM. Hum Brain Mapp 30, 392-402.

Hoeting, J.A., Madigan, D., Raftery, A.E., Volinsky, C.T., 1999. Bayesian model averaging: a tutorial. Stat. Sci. 14, 382–401.

Jansen, B.H., Rit, V.G., 1995. Electroencephalogram and visual evoked potential generation in a mathematical model of coupled cortical columns. Biol Cybern 73, 357-366.

Kass, R., Raftery, A., 1995. Bayes factors. Journal of the American Statistical Association, 773-795.

Kiebel, S.J., David, O., Friston, K.J., 2006. Dynamic causal modelling of evoked responses in EEG/MEG with lead field parameterization. Neuroimage 30, 1273-1284.

Kiebel, S.J., Kloppel, S., Weiskopf, N., Friston, K.J., 2007. Dynamic causal modeling: a generative model of slice timing in fMRI. Neuroimage 34, 1487-1496.

Kiebel, S.J., Garrido, M.I., Moran, R.J., Friston, K.J., 2008. Dynamic causal modelling for EEG and MEG. Cogn Neurodyn 2, 121-136.

Kumar, S., Stephan, K.E., Warren, J.D., Friston, K.J., Griffiths, T.D., 2007. Hierarchical processing of auditory objects in humans. PLoS Comput Biol 3, e100.

Leff, A.P., Schofield, T.M., Stephan, K.E., Crinion, J.T., Friston, K.J., Price, C.J., 2008. The cortical dynamics of intelligible speech. J Neurosci 28, 13209-13215.

Marreiros, A.C., Kiebel, S.J., Friston, K.J., 2008a. Dynamic causal modelling for fMRI: a two-state model. Neuroimage 39, 269-278.

Marreiros, A.C., Daunizeau, J., Kiebel, S.J., Friston, K.J., 2008b. Population dynamics: variance and the sigmoid activation function. Neuroimage 42, 147-157.

Marreiros, A.C., Kiebel, S.J., Daunizeau, J., Harrison, L.M., Friston, K.J., 2009. Population dynamics under the Laplace assumption. Neuroimage 44, 701-714.

Moran, R.J., Stephan, K.E., Seidenbecher, T., Pape, H.C., Dolan, R.J., Friston, K.J., 2009. Dynamic causal models of steady-state responses. Neuroimage 44, 796-811.

Opitz, B., Rinne, T., Mecklinger, A., von Cramon, D.Y., Schroger, E., 2002. Differential contribution of frontal and temporal cortices to auditory change detection: fMRI and ERP results. Neuroimage 15, 167-174.

Penny, W.D., Stephan, K.E., Mechelli, A., Friston, K.J., 2004. Comparing dynamic causal models. Neuroimage 22, 1157-1172.

Penny, W.D., Litvak, V., Fuentemilla, L. Duzel, E., Friston, K., 2009. Dynamic Causal Models for Phase Coupling. J Neurosci Methods, 183(1):19-30.

Penny, W.D., Stephan, K.E., Daunizeau, J., Joao, M., Friston, K., Schofield, T., Leff, A.P., 2010. Comparing Families of Dynamic Causal Models. PLoS Computational Biology, in press.

Posner, M.I., Sheese, B.E., Odludas, Y., Tang, Y., 2006. Analyzing and shaping human attentional networks. Neural Netw 19, 1422-1429.

Raftery, A.E., 1995. Bayesian model selection in social research. Sociological Methodology 1995, Vol 25, 111-163.

Schwarz, G.E., 1978. "Estimating the dimension of a model". Annals of Statistics 6 (2): 461–464.

Smith, A.P., Stephan, K.E., Rugg, M.D., Dolan, R.J., 2006. Task and content modulate amygdala-hippocampal connectivity in emotional retrieval. Neuron 49, 631-638.

Stephan, K.E., Penny, W.D., Marshall, J.C., Fink, G.R., Friston, K.J., 2005. Investigating the functional role of callosal connections with dynamic causal models. Ann N Y Acad Sci 1064, 16-36.

Stephan, K.E., Baldeweg, T., Friston, K.J., 2006. Synaptic plasticity and dysconnection in schizophrenia. Biol Psychiatry 59, 929-939.

Stephan, K.E., Harrison, L.M., Kiebel, S.J., David, O., Penny, W.D., Friston, K.J., 2007a. Dynamic causal models of neural system dynamics: current state and future extensions. J Biosci 32, 129-144.

Stephan, K.E., Marshall, J.C., Penny, W.D., Friston, K.J., Fink, G.R., 2007b. Interhemispheric integration of visual processing during task-driven lateralization. J Neurosci 27, 3512-3522.

Stephan, K.E., Weiskopf, N., Drysdale, P.M., Robinson, P.A., Friston, K.J., 2007c. Comparing hemodynamic models with DCM. Neuroimage 38, 387-401.

Stephan, K.E., Kasper, L., Harrison, L.M., Daunizeau, J., den Ouden, H.E., Breakspear, M., Friston, K.J., 2008. Nonlinear dynamic causal models for fMRI. Neuroimage 42, 649-662.

Stephan, K.E., Penny, W.D., Daunizeau, J., Moran, R.J., Friston, K.J., 2009. Bayesian model selection for group studies. Neuroimage 46: 1004-1017.

Stephan, K.E., Penny, W.D., Moran, R.J., Den Ouden, H.E., Daunizeau, J., Friston, K.J., 2010. Ten simple rules for dynamic causal modelling. Neuroimage 49: 3099-3109.