Spatio-temporal relationships between BOLD and MEG signals at rest or during visuospatial attention

Submission No:

1910 

Submission Type:

Abstract Submission 

Authors:

Chiara Favaretto1, Sara Spadone2, Stefania Della Penna2, Angelo Cenedese1, Maurizio Corbetta1

Institutions:

1University of Padova, Padova, Italy, 2D'Annunzio University of Chieti-Pescara, Chieti, Italy

E-Poster

Introduction:

The relationship between fMRI and MEG signals between different cortical regions (functional connectivity, FC) has been extensively analyzed in the resting state (De Pasquale et al 2010; Brookes et al 2011; Hipp et al 2012). Much less is known about FC modulations from rest to task states, and how they appear respectively in these two imaging modalities. Previously we have shown task-specific alterations of FC in fMRI during a visuospatial attention task (Spadone et al., PNAS, 2015). Specifically, decrements of resting correlation in visual areas were coupled with increments of correlation between visual and dorsal attention regions. Here, we compared fMRI with band-limited power (BLP) correlation obtained with MEG on the same group of subjects. Aim (i) is to measure frequency specific task-related FC modulations in MEG. Aim (ii) is to compare fMRI- and MEG-FC modulations (task-rest).

Methods:

BOLD and MEG time series were recorded in 16 right-handed healthy young subjects during simple fixation (rest state, RS) and task. The task fully described in (Spadone, 2015) involves either maintaining peripheral attention to a stream of sensory stimuli or shifting attention to a different stream simultaneously presented in the opposite visual field. Dorsal attention network (DAN) and occipital visual (VIS) network regions (14 ROIs altogether) localized with fMRI were used for the analysis. Using distributed source modeling, we estimated band limited powers (BLP) at six frequency bands (δ, θ, α, β, γ1, γ2) in these regions. BOLD and BLP functional connectivity (FC) maps were computed as the Pearson correlation coefficient (z Fisher-transformed) between the time series of each ROIs. For aim (i), we used statistical tests (ANOVA, t-test) to investigate significant interaction of factors such as frequency, hemisphere, and network with rest and task condition on BLP-FC. Aim (ii) was investigated through statistical tests on the Pearson's correlation among BOLD- and BLP-FC, applied to evaluate significant relationships among them. In addition, we used the concepts from information and graph theory, such as density and Hamming distance, to define specific measures used to quantify differences among FC matrices, from a network-structural point of view.

Results:

MEG BLP-FC decreased during visuospatial attention, especially in the alpha and beta bands (frequency X condition (F = 35.99; pval < 0.0001). Interestingly, this decrement involved both the visual network, in which fMRI FC was also weaker, and VIS-DAN FC, which was stronger in fMRI. We also observed that some specific connections (e.g. FEF to V3AV7, SPL to MT) between networks were increased in task as compared to rest in the gamma band consistently with increased observed with fMRI (Fig. 1).
These findings indicate that task-rest fMRI FC modulations have variable relationships with corresponding MEG-BLP FC modulations. This result was confirmed by directly comparing fMRI and MEG ROI-to-ROI correlation matrices. The correlation between FC[MEG] and FC[fMRI] was higher during task than rest, consistently with the synchronization induced by performing a task. However, DAN-VIS FC in fMRI and MEG were negatively correlated in the alpha band (t-test; pval < 0.05), but positively correlated in beta (t-test; pval < 0.05) and gamma bands (t-test; pval < 0.0001) (Fig. 2).
Supporting Image: Fig1.png
Supporting Image: Fig_2.png
 

Conclusions:

Prior work has emphasized a positive relationship between alpha/beta BLP correlation and fMRI-FC in the resting state. This relationship apparently is modulated in a frequency specific manner when entering a visuospatial attention task. Decrements of fMRI-FC correspond to decrements in alpha/beta band BLP correlation, as expected. However, task-related increments of fMRI-FC correspond to both alpha decrements and beta/gamma increments in MEG BLP correlation.

Imaging Methods:

BOLD fMRI 2
MEG 1
Multi-Modal Imaging

Modeling and Analysis Methods:

EEG/MEG Modeling and Analysis

Perception and Attention:

Attention: Visual

Keywords:

Data analysis
FUNCTIONAL MRI
MEG

1|2Indicates the priority used for review