Independence of functional connectivity analysis in fMRI research does not rely on whether seeds are exogenous or endogenous

Abstract

Defining seeds is the first step in assessing functional connectivity between separate brain regions. Circular analysis is often a serious concern when defining ROIs. It was considered as circular analysis that seeds were localized on the basis of activations in fMRI data and the functional connectivity among these seeds was measured by using the same data. Such circularity was reported to contaminate results because the margin of detected activations can be affected by noise in the data. Therefore, using independent data for seed selection was advocated. For example, suppose that datasets A and B are independent. Activations are first detected in dataset A through statistical mapping, and then these activations are used to localize seeds for assessing functional connectivity in dataset B. However, localizing seeds only on the basis of an independent source without referring to the intrinsic features of the fMRI data to be analyzed may be inappropriate for assessing functional connectivity. This study demonstrates how different spatial locations and seed scopes influence the results of functional connectivity in task-state fMRI data when these choices meet independence requirements. The independence of analysis does not rely on whether seeds are exogenous or endogenous for the data to be analyzed. Activation and connectivity analyses have inherently independent natures.

Keywords

• fMRI
• Functional connectivity
• Regions of interest (ROI)
• Pearson's correlation coefficient
• t-test