Effective connectivity modeling of human brain

Three different connectivity models, namely structural (anatomical links), functional (statistical dependencies), and effective (causal interactions) have been introduced to examine the interactions between different regions of the brain. To understand the interactive systems, it is of fundamental importance to distinguish the sender from the receiver, and hence to be able to estimate the direction of the information flow. Effective connectivity methods estimate direct and/or indirect causal relationships between the regions. Five different approaches have been proposed in the literature for functional and effective connectivity modeling of the brain: Covariance analysis, information theory, Granger causality based multivariate autoregressive modeling, dynamic causal modeling (DCM), dynamic Bayesian networks (DBN). These modeling approaches handle data in different domains such as time or frequency. All the methods use the data directly to establish a connectivity model with the exception of DCM that starts this process with a proposal model. Except DBN that is probabilistic, all the methods are deterministic. Most of the methods can produce linear models with only a few (DBN and information theory based methods) also being capable of constructing nonlinear models. Since DBN is the only probabilistic and nonlinear method that can model multivariate relations and it is data driven where no prior model suggestion or user tuned parameters are required for its structure learning, the results of DBN based effective connectivity modeling are presented for two studies: 1) Based on the theory that dyslexia is a disconnection syndrome, connectivity models of EEG data recorded from 31 dyslexic and 25 healthy children during word and pseudoword reading experiments are compared to determine whether there are differences between the two groups (dyslexic and control) and the two conditions (word and pseudoword). 2) To reveal the neural circuits for number perception in human brain and then to investigate the problems in these circuits in children with dyscalculia, connectivity models of fMRI data recorded from 6 healthy and 6 dyscalculic children during approximate and symbolic counting experiments are compared to determine whether there are differences between the two groups (dyscalculic and control
Citation Formats
İ. Ulusoy, “Effective connectivity modeling of human brain,” presented at the Turking Human Brain Mapping Meeting, (21 - 25 Kasım 2017), 2017, Accessed: 00, 2021. [Online]. Available: https://hdl.handle.net/11511/78854.