Resolving Resting State Correlations in Noncompliant Subjects Using Short TE FMRI Data

Molly Gallogly Bright¹, Kevin Murphy¹

Characterization of resting state networks may provide insight into clinical populations. However, patients often exhibit increased motion artifacts, obscuring resting state correlations. Short echo time (TE) fMRI data reflect several noise sources, and may assist in mapping correlated networks in these scenarios. We used a dual-echo sequence to acquire fMRI data at very short (~3ms) and BOLD-weighted TEs. Cued head motion was introduced to simulate noncompliant patient behavior. The short TE data were used as voxelwise noise regressors to correct the BOLD-weighted data. This correction recovered seed correlation maps of the default mode network in data with gross motion artifacts.