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

Goelman, Gadi; Dan, Rotem; Keadan, Tarek

doi:10.1038/s41598-018-34672-5

Cited by 12 publications

(21 citation statements)

References 42 publications

(47 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This study aimed to test whether the connectivity between the insula and the sensorimotor cortex is asymmetric, whether this asymmetry is unique to PD and whether this uniqueness is related to non-motor symptoms. For these studies, we used our novel multivariate analysis method with resting-state functional magnetic resonance imaging (rs-fMRI) data (Goelman et al, 2018(Goelman et al, , 2019(Goelman et al, , 2021Goelman & Dan, 2017). This method infers macroscopic anterograde functional pathways, describing directional coupling among four anatomical regions.…”

Section: Introductionmentioning

confidence: 99%

Asymmetry of the insula‐sensorimotor circuit in Parkinson's disease

Goelman

Dan

Růžička

et al. 2021

Eur J of Neuroscience

Self Cite

View full text Add to dashboard Cite

Patients with Parkinson's disease (PD) experience motor and non‐motor symptoms, suggesting alterations of the motor and/or limbic system or more probably of their communications. We hypothesized that the communication between the insula (part of the limbic system) and sensorimotor cortex in PD is altered and hemispheric asymmetric. Furthermore, that this asymmetry relates to non‐motor symptoms, and specifically, that apathy‐related asymmetry is unique to PD. To test these hypotheses, we used a novel multivariate time‐frequency analysis method applied to resting‐state functional magnetic resonance imaging (MRI) data of 28 controls and 25 participants with PD measured in their OFF medication state. The analysis infers directionality of coupling, that is, afferent or efferent, among four anatomical regions, thus defining directed pathways of information flow, which enables the extension of symmetry measures to include directionality. A major right asymmetry reduction of the dorsal‐posterior insula efferent and a slight bilateral increase of insula afferent pathways were observed in participants with PD versus controls. Between‐group pathways that correlated with mild cognitive impairments combined the central‐executive and default‐mode networks through the right insula. Apathy‐correlated pathways of the posterior insula in participants with PD versus controls exhibited reduced right efferent and increased left afferent. Because apathy scores were comparable between the groups and effects of the other motor and non‐motor symptoms were statistically removed by the analysis, the differences in apathy‐correlated pathways were suggested as unique to PD. These pathways could be predictors in the pre‐symptomatic phase in patients with apathy.

show abstract

Section: Introductionmentioning

confidence: 99%

Asymmetry of the insula‐sensorimotor circuit in Parkinson's disease

Goelman

Dan

Růžička

et al. 2021

Eur J of Neuroscience

Self Cite

View full text Add to dashboard Cite

show abstract

“…In this manuscript, we introduced a functional connectivity method for PD that enables to identify altered processes and the effect of levodopa. The method uses a novel multivariate analysis method to identify the mutual interactions between four BOLD signals, deduces their interaction temporal order, and from it the directed pathway of how information is transferred among them (Goelman & Dan, 2017;Goelman et al, 2018Goelman et al, , 2019. We focused on pathways of the motor system by selecting seeds in this system and by calculating pathways that corresponded to three group-relations: "damaged" that is, pathways that were observed only in controls, "preserved" that is, pathways that were unaffected by the disease and "corrected" that is, pathways that were corrected by levodopa.…”

Section: Discussionmentioning

confidence: 99%

“…This method deduces functional anterograde pathways of information flow among four brain regions. Computer simulations of the Kuramoto model tested its accuracy, and its application to the human brain was demonstrated with resting-state and stimulus-driven fMRI data (Goelman & Dan, 2017;Goelman et al, 2018Goelman et al, , 2019. The choice of resting-state fMRI data for the current study was inspired by previous studies showing for example, disrupted functional integration in cortical striatal loops in PD (Buhmann et al, 2003;Gao & Wu, 2016;Tessitore et al, 2014;Thobois et al, 2000;Wu et al, 2009).…”

mentioning

confidence: 99%

Altered sensorimotor fMRI directed connectivity in Parkinson's disease patients

Goelman

Dan

Růžička

et al. 2020

Eur J of Neuroscience

Self Cite

View full text Add to dashboard Cite

Dopamine depletion in the axons of Parkinson's disease (PD) patients precedes depletion in cell bodies thus proposing that macroscopic connectivity can be used to understand disease mechanism. A novel multivariate functional connectivity analysis, based on high order coherence among four fMRI BOLD signals was applied on resting-state fMRI data of controls and PD patients (OFF and ON medication states) and unidirectional multiple-region pathways in the sensorimotor system were identified. Pathways were classified as "preserved" (unaffected by the disease), "damaged" (not observed in patients) and "corrected" (observed in controls and in PD-ON state). The majority of all pathways were feedforward, most of them with the pattern "S1→M1→SMA." Of these pathways, 67% were "damaged," 28% "preserved," and 5% "corrected." Prefrontal cortex (PFC) afferent and efferent pathways that corresponded to goal directed and habitual activities corresponded to recurrent circuits. Eighty-one percent of habitual afferent had internal cue (i.e., M1→S1→), of them 79% were "damaged" and the rest "preserved." All goal-directed afferent had external cue (i.e., S1→M1→) with third "damaged," third "preserved," and third "corrected." Corrected pathways were initiated in the dorsolateral PFC. Reduced connectivity of the SMA and PFC resulted from reduced sensorimotor afferent to these regions. Reduced sensorimotor internal cues to the PFC resulted with reduced habitual processes. Levodopa effects were for pathways that started in region reach with dopamine receptors. This methodology can enrich understudying of PD mechanisms in other (e.g., the default mode network) systems. K E Y W O R D S connectivity through coherence, goal-directed versus habitual activity, multivariate analysis, Parkinson's disease, resting-state fMRI 1 | INTRODUCTION The original model of Parkinson's disease (PD) postulated that dopamine depletion in the basal ganglia (BG) reduced thalami-cortical connectivity (Alexander et al., 1990; Bergman et al., 1990; DeLong, 1990). Recent evidence from imaging, postmortem, neurochemical, and genetic animal models suggested that the axons of the dopaminergic system were predominantly involved in the early stages of PD (Cheng et al.

show abstract

“…This stands in contrast to a growing body of research showing that the functional connectivity of the brain is dynamic and constantly changing over time [5, 6]. As another example, association measures most commonly used are still based on linear models, while it is well known that neuromonitoring data and brain signal in particular interact nonlinearly [7, 8].…”

Section: Introductionmentioning

confidence: 99%

A Deep Graph Neural Network Architecture for Modelling Spatio-temporal Dynamics in resting-state functional MRI Data

Azevedo

Campbell

Romero-García

et al. 2020

Preprint

View full text Add to dashboard Cite

Resting-state functional magnetic resonance imaging (rs-fMRI) has been successfully employed to understand the organisation of the human brain. For rs-fMRI analysis, the brain is typically parcellated into regions of interest (ROIs) and modelled as a graph where each ROI is a node and pairwise correlation between ROI blood-oxygen-level-dependent (BOLD) time series are edges. Recently, graph neural networks (GNNs) have seen a surge in popularity due to their successes in modelling unstructured relational data. The latest developments with GNNs, however, have not yet been fully exploited for the analysis of rs-fMRI data, particularly with regards to its spatio-temporal dynamics. Herein we present a novel deep neural network architecture, combining both GNNs and temporal convolutional networks (TCNs), which is able to learn from the spatial and temporal components of rs-fMRI data in an end-to-end fashion. In particular, this corresponds to intra-feature learning (i.e., learning temporal dynamics with TCNs) as well as inter-feature learning (i.e., leveraging spatial interactions between ROIs with GNNs). We evaluate our model with an ablation study using 35,159 samples from the UK Biobank rs-fMRI database. We also demonstrate explainability features of our architecture which map to realistic neurobiological insights. We hope our model could lay the groundwork for future deep learning architectures focused on leveraging the inherently and inextricably spatio-temporal nature of rs-fMRI data.

show abstract

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

Cited by 12 publications

References 42 publications

Asymmetry of the insula‐sensorimotor circuit in Parkinson's disease

Asymmetry of the insula‐sensorimotor circuit in Parkinson's disease

Altered sensorimotor fMRI directed connectivity in Parkinson's disease patients

A Deep Graph Neural Network Architecture for Modelling Spatio-temporal Dynamics in resting-state functional MRI Data

Contact Info

Product

Resources

About