On the importance of modeling fMRI transients when estimating effective connectivity: A dynamic causal modeling study using ASL data

Havlíček, Martin; Roebroeck, Alard; Friston, Karl J.; Gardumi, Anna; Ivanov, Dimo; Uludağ, Kâmil

doi:10.1016/j.neuroimage.2017.03.017

Cited by 25 publications

(22 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By saying that, our depth-specific model compartments of MV do not represent only venules, but rather a simplified model of both capillaries and venules, being more weighted towards venules. Note that the input to our model is CBF in arterioles, which can be directly linked to CBF measured with ASL data (Havlicek et al, 2017c). However, for small effect sizes (such as the amplitude of the initial dip (Uludağ, 2010)) or reduced oxygenation values in arteries (e.g.…”

Section: Limitations and Future Prospectsmentioning

confidence: 99%

A dynamical model of the laminar BOLD response

Havlíček

Uludağ

2019

Preprint

Self Cite

View full text Add to dashboard Cite

High-resolution functional magnetic resonance imaging (fMRI) using blood oxygenation dependent level-dependent (BOLD) signal is an increasingly popular tool to non-invasively examine neuronal processes at the mesoscopic level. However, as the BOLD signal stems from hemodynamic changes, its temporal and spatial properties do not match those of the underlying neuronal activity. In particular, the laminar BOLD response (LBR), commonly measured with gradient-echo (GE) MRI sequence, is confounded by non-local changes in deoxygenated hemoglobin and cerebral blood volume propagated within intracortical ascending veins, leading to a unidirectional blurring of the neuronal activity distribution towards the cortical surface. Here, we present a new cortical depth-dependent model of the BOLD response based on the principle of mass conservation, which takes the effect of ascending (and pial) veins on the cortical BOLD responses explicitly into account. It can be used to dynamically model cortical depth profiles of the BOLD signal as a function of various baselineand activity-related physiological parameters for any spatiotemporal distribution of neuronal changes. We demonstrate that the commonly observed spatial increase of LBR is mainly due to baseline blood volume increase towards the surface. In contrast, an occasionally observed local maximum in the LBR (i.e. the so-called "bump") is mainly due to spatially inhomogeneous neuronal changes rather than locally higher baseline blood volume. In addition, we show that the GE-BOLD signal laminar point-spread functions, representing the signal leakage towards the surface, depend on several physiological parameters and on the level of neuronal activity. Furthermore, even in the case of simultaneous neuronal changes at each depth, inter-laminar delays of LBR transients are present due to the ascending vein. In summary, the model provides a conceptual framework for the biophysical interpretation of common experimental observations in high-resolution fMRI data. In the future, the model will allow for deconvolution of the spatiotemporal hemodynamic bias of the LBR and provide an estimate of the underlying laminar excitatory and inhibitory neuronal activity.

show abstract

Section: Limitations and Future Prospectsmentioning

confidence: 99%

A dynamical model of the laminar BOLD response

Havlíček

Uludağ

2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Generative models of neuroimaging (Friston et al, 2003;Harrison et al, 2015;Havlicek et al, 2017;Hinne et al, 2014;Langs et al, 2014) or behavioral (Behrens et al, 2007;Friston et al, 2017;Mathys et al, 2014) data have become important pillars of computational and cognitive neuroscience. This type of analysis has the advantage of inferring putative mechanisms underlying neurophysiological and cognitive processes from neuroimaging and behavioral measurements.…”

Section: Introductionmentioning

confidence: 99%

Variational Bayesian inversion for hierarchical unsupervised generative embedding (HUGE)

et al. 2018

View full text Add to dashboard Cite

A B S T R A C TA recently introduced hierarchical generative model unified the inference of effective connectivity in individual subjects and the unsupervised identification of subgroups defined by connectivity patterns. This hierarchical unsupervised generative embedding (HUGE) approach combined a hierarchical formulation of dynamic causal modelling (DCM) for fMRI with Gaussian mixture models and relied on Markov chain Monte Carlo (MCMC) sampling for inference. While well suited for the inversion of complex hierarchical models, MCMC-based sampling suffers from a computational burden that is prohibitive for many applications.To address this problem, this paper derives an efficient variational Bayesian (VB) inversion scheme for HUGE that simultaneously provides approximations to the posterior distribution over model parameters and to the log model evidence. The face validity of the VB scheme was tested using two synthetic fMRI datasets with known ground truth. Additionally, an empirical fMRI dataset of stroke patients and healthy controls was used to evaluate the practical utility of the method in application to real-world problems.Our analyses demonstrate good performance of our VB scheme, with a marked speed-up of model inversion by two orders of magnitude compared to MCMC, while maintaining a similar level of accuracy. Notably, additional acceleration would be possible if parallel computing techniques were applied. Generally, our VB implementation of HUGE is fast enough to support multi-start procedures for whole-group analyses, a useful strategy to ameliorate problems with local extrema. HUGE thus represents a potentially useful practical solution for an important problem in clinical neuromodeling and computational psychiatry, i.e., the unsupervised detection of subgroups in heterogeneous populations that are defined by effective connectivity.

show abstract

“…In contrast, functional connectivity is based on statistical relationships between the activity of neuronal populations and can be easily estimated from recorded signals. For estimating effective connectivity there are methods like Dynamic Causal Modelling, DCM [4,5], Granger causality [6] and others [7,8,9,10,11,12,13]. Only few methods to infer effective connectivity, however, can deal with large numbers of nodes (40 or more) based on zero-lag correlation only.…”

Section: Introductionmentioning

confidence: 99%

From correlation to causation: Estimating effective connectivity from zero-lag covariances of brain signals

et al. 2018

View full text Add to dashboard Cite

Knowing brain connectivity is of great importance both in basic research and for clinical applications. We are proposing a method to infer directed connectivity from zero-lag covariances of neuronal activity recorded at multiple sites. This allows us to identify causal relations that are reflected in neuronal population activity. To derive our strategy, we assume a generic linear model of interacting continuous variables, the components of which represent the activity of local neuronal populations. The suggested method for inferring connectivity from recorded signals exploits the fact that the covariance matrix derived from the observed activity contains information about the existence, the direction and the sign of connections. Assuming a sparsely coupled network, we disambiguate the underlying causal structure via L1-minimization, which is known to prefer sparse solutions. In general, this method is suited to infer effective connectivity from resting state data of various types. We show that our method is applicable over a broad range of structural parameters regarding network size and connection probability of the network. We also explored parameters affecting its activity dynamics, like the eigenvalue spectrum. Also, based on the simulation of suitable Ornstein-Uhlenbeck processes to model BOLD dynamics, we show that with our method it is possible to estimate directed connectivity from zero-lag covariances derived from such signals. In this study, we consider measurement noise and unobserved nodes as additional confounding factors. Furthermore, we investigate the amount of data required for a reliable estimate. Additionally, we apply the proposed method on full-brain resting-state fast fMRI datasets. The resulting network exhibits a tendency for close-by areas being connected as well as inter-hemispheric connections between corresponding areas. In addition, we found that a surprisingly large fraction of more than one third of all identified connections were of inhibitory nature.

show abstract

On the importance of modeling fMRI transients when estimating effective connectivity: A dynamic causal modeling study using ASL data

Cited by 25 publications

References 62 publications

A dynamical model of the laminar BOLD response

A dynamical model of the laminar BOLD response

Variational Bayesian inversion for hierarchical unsupervised generative embedding (HUGE)

From correlation to causation: Estimating effective connectivity from zero-lag covariances of brain signals

Contact Info

Product

Resources

About