The Reliability of Functional Brain Connectivity Measured with MRI

Resting-state functional connectivity MRI (rs-fcMRI), has emerged as a key technique for noninvasively mapping the functional organization of the brain and is poised to have a significant clinical impact, since it allows for the characterization of a large number of functionally relevant circuits without placing any specific task demands on the subject. It also holds great promise for determining the neuro-pathophysiological basis for many mental and neurological disorders.
While it is relatively easy to obtain maps of functional connectivity from resting-state data, a critical problem is that individual differences in rs-fcMRI can be significantly affected by subtle differences in subject motion and physiological noise, even after using current motion and noise correction techniques. For example, recent studies have shown that even small amounts of subject motion can result in correlated signal variations throughout the brain that mimic individual, group, or developmental differences in functional connectivity. Similarly, variations in respiration rate and depth can mimic changes in functional connectivity. As shown in our preliminary data, current physiological noise correction techniques are not accurate enough to fully remove these artifactual signal changes. The confounds from motion and physiological noise present a significant barrier towards using rs-fcMRI for a better understanding of mental or neurological disorders and towards more widespread clinical use – a situation where subject motion and variations in physiology are likely.
The primary goal of this project is to minimize the influence of subject head motion and physiological noise in order to improve the accuracy and consistency of mapping functional brain connectivity. Two complementary strategies are proposed to achieve this goal: 1) the development of a novel MRI acquisition scheme that is more robust to motion and physiological fluctuations; and 2) the development of new processing methods to better model and account for motion and respiration induced artifacts.