B1030
Title: Rethinking artifact removal in functional MRI from a statistical perspective
Authors: Amanda Mejia - Indiana University (United States) [presenting]
Abstract: "Scrubbing" or removal of functional MRI (fMRI) volumes potentially contaminated with artefacts is common practice in fMRI analysis. Scrubbing is most often performed based on measures of subject head motion. However, this practice has become increasingly problematic for a number of reasons. These include a lack of adaptiveness to improved regression-based data denoising techniques; poor generalizability to faster multi-band acquisitions; and over-aggressive removal of volumes with more stringent motion thresholds, often leading to the exclusion of half or more of all sessions. Alternatively, statistical approaches based on abnormalities in the fMRI time series may address many of these limitations and achieve greater sensitivity and specificity. Data-driven alternatives to motion scrubbing are discussed and their potential for dramatically increasing sample sizes in fMRI analysis is illustrated.
