Quantifying the effects of noise variance level on functional connectivity in fMRI under different tasks or conditions: a simulation study

Motivation: In fMRI, the interaction between the dynamics of different brain regions are commonly quantified by correlation-based techniques via (static) functional connectivity (FC)[1], or dynamic FC (DFC) analyses[2]. Goal(s): We aim to quantify the effects of task/condition-related signal variance to FC and DFC. Approach: We perform simulations of fMRI time-signals from different brain regions, compute FC and DFC under different task/conditions and various noise- and signal-variance levels. Results: Significant correlation was found across realizations, between FC and the standard deviations of fMRI time-signals; also, significant correlation across realizations was found between task/condition-related FC-difference and condition-related standard-deviation-difference; this issue was less significant in the DFC. Impact: This study reveals that caution should be taken while applying and interpreting (static) FC analyses of fMRI signals due to noise inflating FC; (dynamic) DFC analysis, capturing the dynamics of the variability, partially mitigates the inflation of FC.