Structural-Spectral Dynamic Causal Modelling for Resting-State fMRI
Poster No:
1367
Submission Type:
Abstract Submission
Authors:
Matthew Greaves1, Leonardo Novelli2, Adeel Razi3
Institutions:
1Monash University, Melbourne, Victoria, 2Monash University, Clayton, Victoria, 3Monash University, Melbourne, Australia
First Author:
Co-Author(s):
Introduction:
We introduce a new dynamic causal model (DCM) for resting-state functional MRI (fMRI) that utilises structural connectivity to characterise power spectra of endogenous neuronal fluctuations. Research indicates a relationship between a brain region's structural connections (i.e., its degree) and the power spectrum of its neuronal activity, typically exhibiting a power-law (scale-free) distribution (Baria et al., 2013; Fallon et al., 2020; Lee & Xue, 2017). Regions with a higher degree exhibit higher power at lower frequencies (i.e., predominant slower fluctuations), which may be due to the role hubs play in coordinating widespread activity across the brain (He, 2014; He et al., 2010). Noting this relationship, we introduce a biologically plausible generative model in which the structural connectivity degree of each brain region characterises the power spectrum of its (random) neuronal fluctuations.
Methods:
Spectral DCM models neuronal (state) noise with generic 1/f spectra, which characterises fluctuations in systems that are at a nonequilibrium steady state (Friston et al., 2014). Spectral DCM parameterises neuronal fluctuations using a power-law form for their power spectra. Under this model, amplitude and exponent parameters control the shape of these power spectra. In the proposed generative model, we provide a new forward model for spectral DCM that utilises the structural connectome. In this model, the degree of each brain region's (normalised) structural connectivity is assumed to have a linear mapping with the exponent parameter of the power spectra. To provide face validation of the proposed structural-spectral DCM, we performed both in silico and empirical illustrations involving a cortico-subcortical effective connectivity network. Thus, we evaluated the performance of the structural-spectral DCM in capturing ground-truth effective connectivity and in describing effective connectivity in a network with substantial interregional differences in terms of the power of low frequency fluctuations (Fallon et al., 2020; He et al., 2010).
Results:
The in silico analyses showed that new structural-spectral DCM was superior in capturing the power spectra of intrinsic neural fluctuations, as compared to the conventional spectral DCM. The model successfully identified hub regions and their associated lower frequency fluctuations, agreeing with the existing evidence about brain network dynamics (e.g., Baria et al., 2013; Fallon et al., 2020). Likewise, this approach was more accurate than spectral DCM in recovering ground-truth effective connectivity. The empirical illustration demonstrated higher model-evidence for the new DCM relative to the spectral DCM.
Conclusions:
The structural-spectral DCM introduced here represents an important advance in characterising endogenous neuronal activity by furnishing a biologically-plausible generative model of their (structurally-informed) power spectra. Although previous research has integrated structural connectivity into DCM via the introduction of empirical priors (e.g., Sokolov et al., 2019, 2020; Stephan et al., 2009), the proposed DCM gracefully fused structural connectivity into the generative model. The mechanistic understanding of how structure shapes the intrinsic functional dynamics afforded by this new DCM will have broad implications for probing pathophysiology of various brain disorders in which power spectra of neural activity is altered (e.g., Trakoshis et al., 2020).
Modeling and Analysis Methods:
Bayesian Modeling 1
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling
Methods Development 2
Task-Independent and Resting-State Analysis
Keywords:
Computational Neuroscience
FUNCTIONAL MRI
STRUCTURAL MRI
1|2Indicates the priority used for review
Provide references using author date format
Fallon, J. (2020). Timescales of spontaneous fMRI fluctuations relate to structural connectivity in the brain. Network Neuroscience, 4(3), 788–806. https://doi.org/10.1162/netn_a_00151
Friston, K. J. (2014). A DCM for resting state fMRI. NeuroImage, 94, 396–407. https://doi.org/10.1016/j.neuroimage.2013.12.009
He, B. J. (2014). Scale-free brain activity: Past, present, and future. Trends in Cognitive Sciences, 18(9), 480–487. https://doi.org/10.1016/j.tics.2014.04.003
He, B. J. (2010). The Temporal Structures and Functional Significance of Scale-free Brain Activity. Neuron, 66(3), 353–369. https://doi.org/10.1016/j.neuron.2010.04.020
Lee, T.-W. (2017). Linking graph features of anatomical architecture to regional brain activity: A multi-modal MRI study. Neuroscience Letters, 651, 123–127. https://doi.org/10.1016/j.neulet.2017.05.005
Sokolov, A. A. (2019). Linking structural and effective brain connectivity: Structurally informed Parametric Empirical Bayes (si-PEB). Brain Structure and Function, 224(1), 205–217. https://doi.org/10.1007/s00429-018-1760-8
Sokolov, A. A. (2020). Asymmetric high-order anatomical brain connectivity sculpts effective connectivity. Network Neuroscience, 4(3), 871–890. https://doi.org/10.1162/netn_a_00150
Stephan, K. E. (2009). Tractography-based priors for dynamic causal models. NeuroImage, 47(4), 1628–1638. https://doi.org/10.1016/j.neuroimage.2009.05.096
Trakoshis, S. (2020). Intrinsic excitation-inhibition imbalance affects medial prefrontal cortex differently in autistic men versus women. eLife, 9, e55684. https://doi.org/10.7554/eLife.55684