Characterizing directed functional pathways in the visual system by multivariate nonlinear coherence of fMRI data

• Published in: Scientific reports Vol. 8; no. 1; pp. 16362 - 14
• Main Authors: Goelman, Gadi, Dan, Rotem, Keadan, Tarek
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 05.11.2018; Nature Publishing Group
• Subjects: 59; 59/36; 631/378/2613/2614; 631/553/2701; 631/553/2712; Functional magnetic resonance imaging; Hemispheric laterality; Humanities and Social Sciences; Medical imaging; multidisciplinary; Neural networks; Neuroimaging; Science; Science (multidisciplinary); Temporal lobe; Visual cortex; Visual system; Resting-state fMRI Data; Feedforward Pathway; Extrastriate Cortex; Vent System; Dorsal System
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-018-34672-5

A multivariate measure of directed functional connectivity is used with resting-state fMRI data of 40 healthy subjects to identify directed pathways of signal progression in the human visual system. The method utilizes 4-nodes networks of mutual interacted BOLD signals to obtains their temporal hierarchy and functional connectivity. Patterns of signal progression were defined at frequency windows by appealing to a hierarchy based upon phase differences, and their significance was assessed by permutation testing. Assuming consistent phase relationship between neuronal and fMRI signals and unidirectional coupling, we were able to characterize directed pathways in the visual system. The ventral and dorsal systems were found to have different functional organizations. The dorsal system, particularly of the left hemisphere, had numerous feedforward pathways connecting the striate and extrastriate cortices with non-visual regions. The ventral system had fewer pathways primarily of two types: (1) feedback pathways initiated in the fusiform gyrus that were either confined to the striate and the extrastriate cortices or connected to the temporal cortex, (2) feedforward pathways initiated in V2, excluded the striate cortex, and connected to non-visual regions. The multivariate measure demonstrated higher specificity than bivariate (pairwise) measure. The analysis can be applied to other neuroimaging and electrophysiological data.