Confounds in Functional Magnetic Resonance Imaging—Part I: Mechanisms

In this review, the first in a three-part series, we consider the origins of confounding in fMRI. We review the signal properties of an fMRI experiment and the intended sensitivity to neurovascular coupling, the hemodynamic response function, and the spatial–temporal biases imposed by the heterogeneous anatomy of the vascular system. We then explore the various mechanisms that can leave nuisance signals, often termed physiologic noise, in fMRI data. All mechanisms having a biological dependence are considered, including mechanical effects from head motion and respiration, blood gases, brain motion from the cardiac pressure wave, blood pressure, and scanner instabilities that couple with the participant's body.