Effective Connectivity Modeling for fMRI: Six Issues and Possible Solutions Using Linear Dynamic Systems

Brain Imaging and Modeling Section, National Institute on Deafness and Other Communication Disorders, National Institutes of Health Bethesda, MD, USA.
Frontiers in Systems Neuroscience 01/2011; 5:104. DOI: 10.3389/fnsys.2011.00104
Source: PubMed

ABSTRACT Analysis of directionally specific or causal interactions between regions in functional magnetic resonance imaging (fMRI) data has proliferated. Here we identify six issues with existing effective connectivity methods that need to be addressed. The issues are discussed within the framework of linear dynamic systems for fMRI (LDSf). The first concerns the use of deterministic models to identify inter-regional effective connectivity. We show that deterministic dynamics are incapable of identifying the trial-to-trial variability typically investigated as the marker of connectivity while stochastic models can capture this variability. The second concerns the simplistic (constant) connectivity modeled by most methods. Connectivity parameters of the LDSf model can vary at the same timescale as the input data. Further, extending LDSf to mixtures of multiple models provides more robust connectivity variation. The third concerns the correct identification of the network itself including the number and anatomical origin of the network nodes. Augmentation of the LDSf state space can identify additional nodes of a network. The fourth concerns the locus of the signal used as a "node" in a network. A novel extension LDSf incorporating sparse canonical correlations can select most relevant voxels from an anatomically defined region based on connectivity. The fifth concerns connection interpretation. Individual parameter differences have received most attention. We present alternative network descriptors of connectivity changes which consider the whole network. The sixth concerns the temporal resolution of fMRI data relative to the timescale of the inter-regional interactions in the brain. LDSf includes an "instantaneous" connection term to capture connectivity occurring at timescales faster than the data resolution. The LDS framework can also be extended to statistically combine fMRI and EEG data. The LDSf framework is a promising foundation for effective connectivity analysis.

Download full-text


Available from: Kewei Chen, Jul 29, 2015
  • Source
    • "Prior attempts have relied on Granger causality [Brovelli et al., 2004] and dynamic causal modeling [Friston et al., 2003; McIntosh and Gonzalez-Lima, 1994]. However, these observational methods rely on statistical covariance [Smith et al., 2011] as opposed to interventional empiric testing. Combining transcranial magnetic stimulation (TMS) with electroencephalography (EEG), magnetoencephalography (MEG), or functional MRI (fMRI) provides a more empiric effective connectivity assessment [Massimini et al., 2005], but this method also has limitations related to the difficulty of inferring intracranial neural measurements from extracranial stimulation using assumptions of EEG/ MEG source modeling or indirect measures of neural activity with fMRI. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The role of cortical connectivity in brain function and pathology is increasingly being recognized. While in vivo magnetic resonance imaging studies have provided important insights into anatomical and functional connectivity, these methodologies are limited in their ability to detect electrophysiological activity and the causal relationships that underlie effective connectivity. Here, we describe results of cortico-cortical evoked potential (CCEP) mapping using single pulse electrical stimulation in 25 patients undergoing seizure monitoring with subdural electrode arrays. Mapping was performed by stimulating adjacent electrode pairs and recording CCEPs from the remainder of the electrode array. CCEPs reliably revealed functional networks and showed an inverse relationship to distance between sites. Coregistration to Brodmann areas (BA) permitted group analysis. Connections were frequently directional with 43% of early responses and 50% of late responses of connections reflecting relative dominance of incoming or outgoing connections. The most consistent connections were seen as outgoing from motor cortex, BA6–BA9, somatosensory (SS) cortex, anterior cingulate cortex, and Broca's area. Network topology revealed motor, SS, and premotor cortices along with BA9 and BA10 and language areas to serve as hubs for cortical connections. BA20 and BA39 demonstrated the most consistent dominance of outdegree connections, while BA5, BA7, auditory cortex, and anterior cingulum demonstrated relatively greater indegree. This multicenter, large-scale, directional study of local and long-range cortical connectivity using direct recordings from awake, humans will aid the interpretation of noninvasive functional connectome studies. Hum Brain Mapp, 2014. © 2014 Wiley Periodicals, Inc.
    Human Brain Mapping 07/2014; 35(12). DOI:10.1002/hbm.22581 · 6.92 Impact Factor
  • Source
    • "Commonly studied resting-state networks (RSNs) include the sensorimotor, visual, default mode, and executive control networks (Beckmann et al., 2005; Damoiseaux et al., 2006; Raichle et al., 2001; Seeley et al., 2007). From TB-FMRI, a variety of inter-regional FC measures can be derived such as with structural equation modeling (McIntosh and Gonzalez-Lima, 1994), vector autoregressive modeling (Goebel et al., 2003), structural vector autoregressive analysis (Chen et al., 2011), psychophysiological interactions (Friston et al., 1997), dynamic causal modeling (Friston et al., 2003), and switching linear dynamic system for fMRI (Smith et al., 2011). While the models underlying these varied approaches differ considerably, the result is a measure of connectedness between regions. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We present a suite of software tools for facilitating the combination of FMRI and diffusion-based tractography from a network-focused point of view. The programs have been designed for investigating functionally-derived GM networks and related structural WM networks. The software comprises the Functional And Tractographic Connectivity Analysis Toolbox (FATCAT), now freely distributed with AFNI. This toolbox supports common file formats and has been designed to integrate as easily as possible with existing standard FMRI pipelines and diffusion software, such as AFNI, FSL and TrackVis. The programs are efficient, run by commandline for facilitating group processing and produce several visualizable outputs. Here, we present the programs and their underlying methods, and we also provide a test example of resting state FMRI analysis combined with tractography. Tractography results are compared with existing methods, showing significantly reduced runtime and generally similar connectivity, but with important differences such as more circumscribed tract regions and a more physiologically identifiable paths produced between several ROI pairs. Currently, FATCAT uses only DT-based tractography (one direction per voxel), but higher order models will soon be included.
    08/2013; 3(5). DOI:10.1089/brain.2013.0154
  • Source
    • "B t , G t ) can be represented as a weighted combination of the given A i (resp. B i , G i ) matrices (Smith et al., 2011b). "
    [Show abstract] [Hide abstract]
    ABSTRACT: The number and variety of connectivity estimation methods is likely to continue to grow over the coming decade. Comparisons between methods are necessary to prune this growth to only the most accurate and robust methods. However, the nature of connectivity is elusive with different methods potentially attempting to identify different aspects of connectivity. Commonalities of connectivity definitions across methods upon which base direct comparisons can be difficult to derive. Here, we explicitly define "effective connectivity" using a common set of observation and state equations that are appropriate for three connectivity methods: dynamic causal modeling (DCM), multivariate autoregressive modeling (MAR), and switching linear dynamic systems for fMRI (sLDSf). In addition while deriving this set, we show how many other popular functional and effective connectivity methods are actually simplifications of these equations. We discuss implications of these connections for the practice of using one method to simulate data for another method. After mathematically connecting the three effective connectivity methods, simulated fMRI data with varying numbers of regions and task conditions is generated from the common equation. This simulated data explicitly contains the type of the connectivity that the three models were intended to identify. Each method is applied to the simulated data sets and the accuracy of parameter identification is analyzed. All methods perform above chance levels at identifying correct connectivity parameters. The sLDSf method was superior in parameter estimation accuracy to both DCM and MAR for all types of comparisons.
    Frontiers in Neuroscience 05/2013; 7:70. DOI:10.3389/fnins.2013.00070 · 3.70 Impact Factor
Show more