A Multi-Subject, Dynamic Bayesian Networks (DBNS) Framework for Brain Effective Connectivity

Abstract

As dynamic connectivity is shown essential for normal brain function and is disrupted in disease, it is critical to develop models for inferring brain effective connectivity from non-invasive (e.g., fMRI) data. Increasingly, (dynamic) Bayesian network (BNs) have been suggested for this purpose due to their exibility and suitability. However, ultimately extrapolating BN results from one subject to an entire population rst requires methods meaningfully addressing inter-subject, within-group variability. Here we explore two group analysis approaches in fMRI using DBNs: one is to construct a group network based on a common structure assumption across individuals, and the other is to identify signi cant structure features by examining DBNs individually-trained. By investigating real fMRI data from Parkinsons Disease (PD) and normal subjects performing a motor task at three progressive levels of dif culty, we noted that both methods detected statistically signi cant, biologically plausible connectivity between task-related region-of-interest (ROIs) that differed between the PD and normal subjects. However, the second approach was more sensitive, nding more features that were also consistent with prior neuroscience knowledge. Determining highly reproducible DBN nodes/edges across subjects seems promising for inferring altered functional connectivity within a group.

Bayesian network models in brain functional connectivity analysis

“…The development of efficient fMRI data preprocessing methods and accessible software’s made possible a more reliable execution of group connectivity analyses. Pioneering studies applying BN to functional connectivity analysis of fMRI were published by Labatut and colleagues [14], followed by others [15, 16] [9, 17, 18]. Recently, BN and its methodological extensions were suggested to be useful in inferring causal relationships between activations by Glymour and colleagues [7].…”

“…Individual-structure (IS) approach : it learns individual networks for each subject separately, and performs group-analysis on these individual networks [9]. Thus, for each subject we learn a BN structure with a score.…”

“…Although much progress has been made in the field, the interdisciplinary dialogue between statistician, computer scientist, physician and psychologist only began recently. For instance, although investigations have successfully applied Bayesian network (BN) and probabilistic graphical models in functional connectivity analysis [2, 7], suggesting a large venue of opportunities to be explored, few functional connectivity studies systematically addressed the use of BN modeling in magnetic resonance imaging (fMRI) of a large number of subjects executing a cognitive task [8, 9]. …”

Dynamic Bayesian network modeling of fMRI: A comparison of group-analysis methods

“…At the other extreme, the "individual-structure" (IS) approach learns a network for each subject separately and then performs group analysis on the individually learned networks (Goncalves et al, 2001;Li et al, 2007). The IS approach is consistent with the concept of functional degeneracy, i.e., "the ability of elements that are structurally different to perform the same function or yield the same output" (Edelman and Gally, 2001), or more plainly, "there are multiple ways of completing the same task" .…”

Multi-Scale Information, Network, Causality, and Dynamics: Mathematical Computation and Bayesian Inference to Cognitive Neuroscience and Aging

“…The functional structures captured on two fMRI datasets are consistent with the previous literature findings and more accurate than those identified by BN. Li et al [84] aimed to extrapolate BN results from one subject to an entire population while addressing inter-subject, within-group variability. The authors explored two group analysis approaches in fMRI using DBNs: constructing a group network based on a common structure assumption across individuals, and identifying significant structure features by examining DBNs individually-trained.…”