Functional connectivity in the rat at 11.7 T: Impact of physiological noise in resting state fMRI

Resting state functional MRI (rs-fMRI) of the brain has the potential to elicit networks of functional connectivity and to reveal changes thereof in animal models of neurological disorders. In the present study, we investigate the contribution of physiological noise and its impact on assessment of functional connectivity in rs-fMRI of medetomidine sedated, spontaneously breathing rats at ultrahigh field of 11.7 Tesla. We employed gradient echo planar imaging (EPI) with repetition times of 3 s and used simultaneous recordings of physiological parameters. A model of linear regression was applied to quantify the amount of BOLD fMRI signal fluctuations attributable to physiological sources. Our results indicate that physiological noise - mainly originating from the respiratory cycle -dominates the rs-fMRI time course in the form of spatially complex correlation patterns. As a consequence, these physiological fluctuations introduce severe artifacts into seed-based correlation maps and lead to misinterpretation of corresponding connectivity measures. We demonstrate that a scheme of motion correction and linear regression can significantly reduce physiological noise in the rs-fMRI time course, remove artifacts, and hence improve the reproducibility of functional connectivity assessment. In conclusion, physiological noise can severely compromise functional connectivity MRI (fcMRI) of the rodent at high fields and must be carefully considered in design and interpretation of future studies. Motion correction should be considered the primary strategy for reduction of apparent motion related to respiratory fluctuations. Combined with subsequent regression of physiological confounders, this strategy has proven successful in reducing physiological noise and related artifacts affecting functional connectivity analysis. The proposed new and rigorous protocol now opens the potential of fcMRI to elicit the role of brain connectivity in pathological processes without concerns of confounding contributions from physiological noise. (C) 2010 Elsevier Inc. All rights reserved.